further education settings. Public Health Research 5(8 ... · Pilot trial and process evaluation of...

Willmott, M., Langford, B., Campbell, R., & Fletcher, A. (2017). Pilot trialand process evaluation of a multi-level smoking prevention intervention infurther education settings. Public Health Research, 5(8).https://doi.org/10.3310/phr05080

Publisher's PDF, also known as Version of record

License (if available):Other

Link to published version (if available):10.3310/phr05080

Link to publication record in Explore Bristol ResearchPDF-document

This is the final published version of the article (version of record). It first appeared online via National Institutefor Health Research at https://www.journalslibrary.nihr.ac.uk/phr/phr05080#/full-report. Please refer to anyapplicable terms of use of the publisher.

University of Bristol - Explore Bristol ResearchGeneral rights

This document is made available in accordance with publisher policies. Please cite only the publishedversion using the reference above. Full terms of use are available: http://www.bristol.ac.uk/pure/user-guides/explore-bristol-research/ebr-terms/

https://doi.org/10.3310/phr05080

https://doi.org/10.3310/phr05080

https://research-information.bris.ac.uk/en/publications/pilot-trial-and-process-evaluation-of-a-multilevel-smoking-prevention-intervention-in-further-education-settings(e5d27a13-9c5c-4768-9ca2-2b1556a37097).html

https://research-information.bris.ac.uk/en/publications/pilot-trial-and-process-evaluation-of-a-multilevel-smoking-prevention-intervention-in-further-education-settings(e5d27a13-9c5c-4768-9ca2-2b1556a37097).html

PUBLIC HEALTH RESEARCHVOLUME 5 ISSUE 8 OCTOBER 2017

ISSN 2050-4381

DOI 10.3310/phr05080

Pilot trial and process evaluation of a multilevel smoking prevention intervention in further education settings

Adam Fletcher, Micky Willmott, Rebecca Langford, James White, Ria Poole, Rachel Brown, Honor Young, Graham Moore, Simon Murphy, Julia Townson, William Hollingworth, Rona Campbell and Chris Bonell

Pilot trial and process evaluation of amultilevel smoking prevention interventionin further education settings

Adam Fletcher,1* Micky Willmott,2 Rebecca Langford,2

James White,3 Ria Poole,1 Rachel Brown,1

Honor Young,1 Graham Moore,1 Simon Murphy,1

Julia Townson,3 William Hollingworth,2

Rona Campbell2 and Chris Bonell4

1Centre for the Development and Evaluation of Complex Interventions for PublicHealth Improvement (DECIPHer), School of Social Sciences, Cardiff University,Cardiff, UK

2DECIPHer, School of Social and Community Medicine, University of Bristol,Bristol, UK

3Cardiff Centre for Trials Research, School of Medicine, Cardiff University, Cardiff, UK4Department of Social and Environmental Health Research, London School ofHygiene and Tropical Medicine, London, UK

*Corresponding author

Declared competing interests of authors: Adam Fletcher, Graham Moore and Chris Bonell aremembers of the Public Health Research (PHR) Research Funding Board. Rona Campbell is a member of thePHR Research Funding Board, and reports grants from the University of Bristol during the conduct of thestudy and personal fees from DECIPHer Impact Limited, outside the submitted work. Note that the Centrefor the Development and Evaluation of Complex Interventions for Public Health Improvement (DECIPHer)and DECIPHer Impact Ltd are separate entities, the latter being a not-for-profit organisation.

Disclaimer: This report contains transcripts of interviews conducted in the course of the research andcontains language that may offend some readers.

Published October 2017DOI: 10.3310/phr05080

This report should be referenced as follows:

Fletcher A, Willmott M, Langford R, White J, Poole R, Brown R, et al. Pilot trial and process

evaluation of a multilevel smoking prevention intervention in further education settings.

Public Health Res 2017;5(8).

Public Health Research

ISSN 2050-4381 (Print)

ISSN 2050-439X (Online)

This journal is a member of and subscribes to the principles of the Committee on Publication Ethics (COPE) (www.publicationethics.org/).

Editorial contact: [email protected]

The full PHR archive is freely available to view online at www.journalslibrary.nihr.ac.uk/phr. Print-on-demand copies can be purchased from thereport pages of the NIHR Journals Library website: www.journalslibrary.nihr.ac.uk

Criteria for inclusion in the Public Health Research journalReports are published in Public Health Research (PHR) if (1) they have resulted from work for the PHR programme, and (2) they are of asufficiently high scientific quality as assessed by the reviewers and editors.

Reviews in Public Health Research are termed ‘systematic’ when the account of the search appraisal and synthesis methods (tominimise biases and random errors) would, in theory, permit the replication of the review by others.

PHR programmeThe Public Health Research (PHR) programme, part of the National Institute for Health Research (NIHR), evaluates public health interventions,providing new knowledge on the benefits, costs, acceptability and wider impacts of non-NHS interventions intended to improve the healthof the public and reduce inequalities in health. The scope of the programme is multi-disciplinary and broad, covering a range of interventionsthat improve public health. The Public Health Research programme also complements the NIHR Health Technology Assessment programmewhich has a growing portfolio evaluating NHS public health interventions.

For more information about the PHR programme please visit the website: http://www.nets.nihr.ac.uk/programmes/phr

This reportThe research reported in this issue of the journal was funded by the PHR programme as project number 13/42/02. The contractual start datewas in July 2014. The final report began editorial review in November 2016 and was accepted for publication in March 2017. The authorshave been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The PHR editors and productionhouse have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on thefinal report document. However, they do not accept liability for damages or losses arising from material published in this report.

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed byauthors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the PHR programme orthe Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the intervieweesare those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the PHR programme orthe Department of Health.

© Queen’s Printer and Controller of HMSO 2017. This work was produced by Fletcher et al. under the terms of a commissioningcontract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research andstudy and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgementis made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should beaddressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre,Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Published by the NIHR Journals Library (www.journalslibrary.nihr.ac.uk), produced by Prepress Projects Ltd, Perth, Scotland(www.prepress-projects.co.uk).

Public Health Research Editor-in-Chief

Professor Martin White Director of Research and Programme Leader, UKCRC Centre for Diet and Activity

NIHR Journals Library Editor-in-Chief

Professor Tom Walley Director, NIHR Evaluation, Trials and Studies and Director of the EME Programme, UK

NIHR Journals Library Editors

Research (CEDAR), MRC Epidemiology Unit, Institute of Metabolic Science, School of Clinical Medicine, University of Cambridge; Visiting Professor, Newcastle University; and Director, NIHR Public Health Research Programme

Professor Ken Stein Chair of HTA and EME Editorial Board and Professor of Public Health, University of Exeter Medical School, UK

Professor Andrée Le May Chair of NIHR Journals Library Editorial Group (HS&DR, PGfAR, PHR journals)

Dr Martin Ashton-Key Consultant in Public Health Medicine/Consultant Advisor, NETSCC, UK

Professor Matthias Beck Chair in Public Sector Management and Subject Leader (Management Group), Queen’s University Management School, Queen’s University Belfast, UK

Dr Tessa Crilly Director, Crystal Blue Consulting Ltd, UK

Dr Eugenia Cronin Senior Scientific Advisor, Wessex Institute, UK

Dr Peter Davidson Director of the NIHR Dissemination Centre, University of Southampton, UK

Ms Tara Lamont Scientific Advisor, NETSCC, UK

Dr Catriona McDaid Senior Research Fellow, York Trials Unit, Department of Health Sciences, University of York, UK

Professor William McGuire Professor of Child Health, Hull York Medical School, University of York, UK

Professor Geoffrey Meads Professor of Wellbeing Research, University of Winchester, UK

Professor John Norrie Chair in Medical Statistics, University of Edinburgh, UK

Professor John Powell Consultant Clinical Adviser, National Institute for Health and Care Excellence (NICE), UK

Professor James Raftery Professor of Health Technology Assessment, Wessex Institute, Faculty of Medicine, University of Southampton, UK

Dr Rob Riemsma Reviews Manager, Kleijnen Systematic Reviews Ltd, UK

Professor Helen Roberts Professor of Child Health Research, UCL Institute of Child Health, UK

Professor Jonathan Ross Professor of Sexual Health and HIV, University Hospital Birmingham, UK

Professor Helen Snooks Professor of Health Services Research, Institute of Life Science, College of Medicine, Swansea University, UK

Professor Jim Thornton Professor of Obstetrics and Gynaecology, Faculty of Medicine and Health Sciences, University of Nottingham, UK

Professor Martin Underwood Director, Warwick Clinical Trials Unit, Warwick Medical School,University of Warwick, UK

Please visit the website for a list of members of the NIHR Journals Library Board: www.journalslibrary.nihr.ac.uk/about/editors

Editorial contact: [email protected]

NIHR Journals Library www.journalslibrary.nihr.ac.uk

Abstract

Pilot trial and process evaluation of a multilevel smokingprevention intervention in further education settings

Adam Fletcher,1* Micky Willmott,2 Rebecca Langford,2 James White,3

Ria Poole,1 Rachel Brown,1 Honor Young,1 Graham Moore,1

Simon Murphy,1 Julia Townson,3 William Hollingworth,2

Rona Campbell2 and Chris Bonell4

1Centre for the Development and Evaluation of Complex Interventions for Public HealthImprovement (DECIPHer), School of Social Sciences, Cardiff University, Cardiff, UK

2DECIPHer, School of Social and Community Medicine, University of Bristol, Bristol, UK3Cardiff Centre for Trials Research, School of Medicine, Cardiff University, Cardiff, UK4Department of Social and Environmental Health Research, London School of Hygiene andTropical Medicine, London, UK

*Corresponding author [email protected]

Background: Preventing smoking uptake among young people is a public health priority. Furthereducation (FE) settings provide access to the majority of 16- to 18-year-olds, but few evaluations ofsmoking prevention interventions have been reported in this context to date.

Objectives: To evaluate the feasibility and acceptability of implementing and trialling a new multilevelsmoking prevention intervention in FE settings.

Design: Pilot cluster randomised controlled trial and process evaluation.

Setting: Six UK FE institutions.

Participants: FE students aged 16–18 years.

Intervention: ‘The Filter FE’ intervention. Staff working on Action on Smoking and Health Wales’‘The Filter’ youth project applied existing staff training, social media and youth work resources in threeintervention settings, compared with three control sites with usual practice. The intervention aimed toprevent smoking uptake by restricting the sale of tobacco to under-18s in local shops, implementingtobacco-free campus policies, training FE staff to deliver smoke-free messages, publicising The Filter youthproject’s online advice and support services, and providing educational youth work activities.

Main outcome measures: (1) The primary outcome assessed was the feasibility and acceptability ofdelivering and trialling the intervention. (2) Qualitative process data were analysed to explore student, staffand intervention team experiences of implementing and trialling the intervention. (3) Primary, secondaryand intermediate (process) outcomes and economic evaluation methods were piloted.

Data sources: New students at participating FE settings were surveyed in September 2014 and followedup in September 2015. Qualitative process data were collected via interviews with FE college managers(n = 5) and the intervention team (n = 6); focus groups with students (n = 11) and staff (n = 5); andobservations of intervention settings. Other data sources were semistructured observations of interventiondelivery, intervention team records, ‘mystery shopper’ audits of local shops and college policy documents.

DOI: 10.3310/phr05080 PUBLIC HEALTH RESEARCH 2017 VOL. 5 NO. 8

© Queen’s Printer and Controller of HMSO 2017. This work was produced by Fletcher et al. under the terms of a commissioning contract issued by the Secretary of State forHealth. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journalsprovided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should beaddressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton SciencePark, Southampton SO16 7NS, UK.

v

Results: The intervention was not delivered as planned at any of the three intervention settings, with noimplementation of some community- and college-level components, and low fidelity of the social mediacomponent across sites. Staff training reached 28 staff and youth work activities were attended by 190students across the three sites (< 10% of all eligible staff and students), with low levels of acceptabilityreported. Implementation was limited by various factors, such as uncertainty about the value of smokingprevention activities in FE colleges, intervention management weaknesses and high turnover ofintervention staff. It was feasible to recruit, randomise and retain FE settings. Prevalence of weeklysmoking at baseline was 20.6% and was 17.2% at follow-up, with low levels of missing data for allpilot outcomes.

Limitations: Only 17% of eligible students participated in baseline and follow-up surveys; therepresentativeness of student and staff focus groups is uncertain.

Conclusions: In this study, FE settings were not a supportive environment for smoking prevention activitiesbecause of their non-interventionist institutional cultures promoting personal responsibility. Weaknesses inintervention management and staff turnover also limited implementation. Managers accept randomisationbut methodological work is required to improve student recruitment and retention rates if trials are to beconducted in FE settings.

Trial registration: Current Controlled Trials ISRCTN19563136.

Funding: This project was funded by the National Institute for Health Research (NIHR) Public HealthResearch programme and will be published in full in Public Health Research; Vol. 5, No. 8. See the NIHRJournals Library website for further project information. It was also funded by the Big Lottery Fund.

ABSTRACT


vi

Contents

List of tables ix

List of figures xi

List of boxes xiii

List of abbreviations xv

Plain English summary xvii

Scientific summary xix

Chapter 1 Introduction 1Youth smoking: a public health priority 1Health improvement in further education settings 1Effective smoking prevention methods and approaches 1‘The Filter FE’ intervention design and logic model 2Public involvement 4Study aim, objectives and research questions 4

Chapter 2 Methods 7Study design: overview 7

Intervention components 7Sampling and recruitment of further education settings 8Randomisation 9Progression criteria 9

Data sources 10Data analysis methods 11

Evaluating participants’ experiences of the process 11Qualitative process data 11Data analysis methods 15

Pilot outcome measures 15Pilot primary and secondary outcome measures 16Pilot intermediate outcome measures 16Data sources 17Statistical analyses 19

Economic analysis 19Trial registration, governance and ethics 19

Chapter 3 Results 21Description of pilot trial sample 21

Flow of participants in the pilot trial 22Student characteristics 22

Progression criteria assessment 25Intervention feasibility and acceptability 25Trial feasibility and acceptability 28

Process evaluation: participants’ experiences 29Attitudes towards smoking in the further education context 29



vii

Barriers to acceptability and implementation 33Implementation of intervention components and contextual variation 36Research methods: feasibility and acceptability 49

Pilot primary and secondary outcomes 52Missing data, prevalence and distribution by arm at baseline 52Missing data, prevalence and distribution by arm at follow-up 53Feasibility of assessing cost-effectiveness 57

Pilot intermediate (outcome) measures 62

Chapter 4 Discussion 69Limitations 69

Student survey limitations 69Limitations with other data collection methods 70Deviations from the protocol 71

Key results and generalisability 71Acceptability 71Reach 72

Implications 73Conclusions and recommendations for further research 74

Acknowledgements 77

References 79

CONTENTS


viii

List of tables

TABLE 1 Participating FE settings and allocation to trial arm 9

TABLE 2 Qualitative process data collected by arm and setting 12

TABLE 3 Participants in semistructured observations of staff training sessions 13

TABLE 4 Participants in semistructured observations of youth work sessions 13

TABLE 5 Summary of categorical baseline demographic characteristics accordingto sample eligibility 22

TABLE 6 Summary of categorical baseline demographic characteristics by trial arm 24

TABLE 7 Pilot primary outcome and categorical secondary outcomes at baseline 53

TABLE 8 Numerical secondary outcomes at baseline 54

TABLE 9 Pilot primary outcome and categorical secondary outcomes at follow-up 55

TABLE 10 Numerical secondary outcomes at follow-up 56

TABLE 11 Categorical EQ-5D-5L items and health service use at baseline 58

TABLE 12 Numerical EQ-5D-5L items and health service use at baseline 60

TABLE 13 Categorical EQ-5D-5L items and health service use at follow-up 61

TABLE 14 Numerical EQ-5D-5L items and health service use at follow-up 63

TABLE 15 Pilot intermediate outcome measures at baseline 64

TABLE 16 Pilot intermediate outcome measures at follow-up 65



ix

List of figures

FIGURE 1 Intervention logic model 3

FIGURE 2 The Consolidated Standards of Reporting Trials diagram 21



xi

List of boxes

BOX 1 Research questions addressing the agreed criteria for progression 10

BOX 2 Research questions to evaluate participant experiences 11

BOX 3 Research questions to evaluate outcome measures 16



xiii

List of abbreviations

A level Advanced level

ALPHA Advice Leading to Public HealthAdvancement

AS level Advanced Subsidiary level

ASH Action on Smoking and Health

AUDIT-C Alcohol Use Disorders IdentificationTest Consumption

BTEC Business and Technology EducationCouncil

CI confidence interval

e-cigarette electronic cigarette

EQ-5D-5L EuroQol-5 Dimensions, 5-levelversion

ESFA European Smoking PreventionFramework Approach

FE further education

GCSE General Certificate of SecondaryEducation

GLS General Lifestyle Survey Overview:A Report on the 2010 GeneralLifestyle Survey

HSI Heaviness of Smoking Index

ICC intracluster correlation

NIHR National Institute for HealthResearch

ONS Office for National Statistics

PHR Public Health Research

RCT randomised controlled trial

RQ research question

SD standard deviation



xv

Plain English summary

Smoking remains the leading cause of ill health and early death in the UK. Preventing young peoplefrom taking up smoking is vital to improve public health. More than 1.5 million 16- to 18-year-olds are

now enrolled in further education (FE) courses in the UK, but there remains very little investment insmoking prevention activities in FE colleges or ‘sixth form’ colleges. However, this is the time when manypeople start to smoke, and the transition to FE itself increases the risk of starting smoking for some youngpeople as they make new friends and are more independent of their parents.

This research evaluated a new smoking prevention project for 16- to 18-year-olds that was delivered inboth general FE colleges and ‘sixth form’ colleges. The smoking prevention project is called ‘The Filter FE’.We evaluated this project over 1 college-year to check if it was delivered as planned and how acceptable(or not) it was with staff and students in different colleges. Three colleges received the project and theother three continued with their normal practice and acted as a ‘control group’.

At the three colleges where the project was delivered, a project manager, staff trainers, social mediaexperts and trained youth workers were deployed to implement a range of new smoking preventionactivities. However, prevention activities were not always implemented as intended, such as the plannedsmoke-free campus policies. Staff training reached a total of 28 staff and youth work activities wereattended by 190 students, although many of them felt that the messages about the harms of smokingwere already well known. It was challenging to integrate existing web-based information, social mediacampaigns and online services with the colleges’ websites and social media. Further evaluation of the FilterFE project is not recommended because of the low levels of acceptability to students and staff.



xvii

Scientific summary

Background

Smoking is a major cause of preventable illness, premature death and health inequalities in the UK. Preventingyoung people from taking up smoking is vital to maintain and accelerate recent declines in smoking rates.Although much research has been undertaken to develop and evaluate school-based prevention interventionstargeting 11- to 15-year-olds, smoking continues to grow rapidly among older youth. With > 1.5 million British16- to 18-year-olds now enrolled in further education (FE) courses, new smoking prevention interventions arerequired that target FE settings (e.g. general FE colleges, ‘sixth form’ colleges attached to secondary schools,etc.). As well as being a period in life when smoking often begins, the transition to FE itself can increase therisk of smoking as young people are exposed to new sources of peer influence and have more independencefrom their parents. However, research evidence about preventing smoking among FE students is sparse, withfew evaluations of smoking prevention interventions in FE colleges to date.

To address this gap, ‘The Filter FE’ intervention and logic model was co-designed by Action on Smoking andHealth (ASH) Wales and the research team to apply the educational, training and social media resources fromASH Wales’ ‘The Filter’ youth project to FE settings in 2014–15. The Filter FE is a novel, multilevel interventiontargeting 16- to 18-year-old students in FE settings, delivered by trained staff working on ASH Wales’The Filter youth project. Informed by systematic reviews of smoking prevention interventions delivered inschools and other settings, the intervention was designed to integrate the following prevention methods andapproaches in FE settings: preventing the sale of tobacco to under-18-year-olds in local shops; implementingtobacco-free campus policies; training FE staff to deliver smoke-free messages and support institutionalchange; publicising The Filter youth project’s online social marketing campaigns, advice and support services;and on-site youth work activities to provide credible educational messages, address norms, and promoteresistance skills, as well as signposting to cessation services. To facilitate scalability and sustainability across UKFE settings (including large institutions), the intervention involves standardised processes and activitiesbalanced with opportunities for a local tailoring of activities.

Study aim, objectives and research questions

The aim of the pilot trial was to evaluate the feasibility and acceptability of implementing and trialling theFilter FE intervention. The study had three objectives.

The first objective was to assess whether or not prespecified feasibility and acceptability criteria were metprior to progressing to a larger, Phase III, trial to examine effectiveness. To meet this objective, data werecollected and analysed to address these research questions (RQs):

1. Did the intervention activities occur as planned in (at least) two out of three intervention settings?2. Were the intervention activities delivered with high fidelity across all settings?3. Was the intervention acceptable to the majority of FE managers, staff, students and the intervention

delivery team?4. Was randomisation acceptable to FE managers?5. Did (at least) two out of three colleges from each of the intervention and control arms continue to

participate in the study at the 1-year follow-up?6. Do student survey response rates suggest that we could recruit and retain at least 70% of new students

in both arms in a subsequent effectiveness trial?



xix

The second objective was to explore the experiences of FE students, staff and the intervention teamregarding the pilot intervention and trial design, including how the logic model, intervention content anddata collection methods could be refined. In order to meet this objective, data were collected and analysedto address the following RQs:

7. What are students’, staff’s and intervention team members’ experiences of the intervention and viewsabout its potential impacts on health?

8. What are the barriers to, and facilitators of, implementation, and how do these vary according tocollege context and/or other factors?

9. Were there any unexpected consequences?10. How acceptable were the data collection methods to students and staff, and do participants think

longer-term follow-up via e-mail or telephone interview would be feasible?11. What resources and partnerships are necessary for a Phase III trial?

The third objective was to pilot primary, secondary and intermediate outcome measures and economicevaluation methods prior to a potential effectiveness trial. It was not an objective of the pilot study toassess intervention effects and the study was not designed or powered to do so. Data were collected andanalysed to address the following RQs:

12. Does the primary outcome measure (smoking weekly or more) have an acceptable completion rate,adequate validity and minimise floor/ceiling effects?

13. Do cotinine concentrations of saliva samples indicate any evidence of response bias between arms inself-reported smoking status?

14. Was it feasible and acceptable to measure all the secondary and intermediate outcomes of interest atbaseline and follow-up?

15. Is it feasible to assess cost-effectiveness using a cost–utility analysis within a Phase III trial?

Methods

A cluster randomised controlled pilot trial and process evaluation was undertaken in six FE settings inWales (purposively sampled to examine delivery and trial methods in a range of institutional contexts) withallocation to the Filter FE intervention (three FE settings) or continuation of normal practice (three FEsettings). The following criteria were used to purposively sample FE settings and stratify the allocation:large FE college campuses (new intake > 500 students) (n = 2), small FE college campuses (new intake< 500 students) (n = 2) and ‘sixth form’ colleges attached to schools (n = 2).

In order to assess the feasibility and acceptability of delivering and trialling the intervention accordingto prespecified criteria (objective 1), we collected a range of quantitative and qualitative data viasemistructured observations of staff training sessions (n = 1 per intervention setting), group-based youthwork sessions (n = 1 per intervention setting) and college websites and social media channels (n = 2 perintervention setting); interviews with FE college managers (n = 5) and the intervention team (n = 6); anddocumentary evidence (e.g. college policies, intervention team records, etc.). The retention of FE settingsand response rates were assessed using student survey data.

To explore participants’ experiences of implementing and trialling the Filter FE intervention (objective 2),qualitative process data were collected via interviews with FE college managers (n = 5) and the interventionteam (n = 6); focus groups with students (n = 11) and staff (n = 5); and semistructured observations ofintervention settings. These qualitative data were transcribed verbatim and analysed using techniquesassociated with thematic content analysis and grounded theory. The coding framework included bothdeductive codes, derived from key RQs and relevant progression criteria, and inductive codes, identifyingother relevant themes emerging from the data.

SCIENTIFIC SUMMARY


xx

In addition to examining intervention and trial feasibility and acceptability, primary, secondary, intermediate(process) outcomes and economic evaluation methods were piloted (objective 3). Surveys of new studentsenrolling at the participating FE settings in September 2014 (baseline) and September 2015 (1-year follow-up)were used to examine the pilot primary (self-reported smoking weekly or more) and secondary outcomemeasures (self-reported lifetime smoking, use of cannabis in the past 30 days, frequent cannabis use, high-riskalcohol use and health-related quality of life). The following additional pilot secondary outcomes for baselinesmokers were also examined: cessation, number of cigarettes per week and nicotine dependence. Informedby the intervention logic model, multiple sources of data were also collected at baseline and follow-up to pilotintermediate (process) outcomes at multiple levels: the restriction of the availability of tobacco in local shopswas assessed via ‘mystery shopper’ audits; changes to the institutional environment and policies were assessedvia structured observations and analysis of college policy documents; students’ knowledge, norm and social/situational self-efficacy and resistance skills were assessed via the student survey. Potential economic analysesmethods were assessed, including the use of EuroQol-5 Dimensions, 5-level version (EQ-5D-5L) health-relatedquality-of-life measure. It was not feasible to collect saliva samples from students to assess the validity ofself-reported smoking status at follow-up.

Results

The intervention was not delivered in full at any of the three intervention settings, with no implementation ofsome community- and college-level components, and low fidelity of the social media component across sites.The staff training reached a total of 28 staff and youth work activities were attended by 190 students acrossthe three sites (< 10% of all staff and students). Lower than intended recruitment to these activities waslargely the result of lack of demand from staff at intervention settings and, although those who did attendwere observed to be engaged, low levels of acceptability were reported across FE sites. The interventionteam reported additional challenges to recruitment because of the short lead-in time prior to implementationand high intervention-team staff turnover during the pilot study. The process evaluation also found thatplanned institutional policy review activities did not occur at any of the sites, with limited evidence ofchanges to smoking policies post intervention. This was, again, associated with limited preparation time forintervention delivery as well as issues relating to the management of intervention, which also impacted onlimited community-level activities targeting local shops.

Six colleges were randomised into the two trial arms and all were retained at the 1-year follow-up.Recruitment and retention of students was challenging, despite the use of the multiple methods andincentives. In September 2014, 1320 students out of an estimated total sample of 2363 participated in thebaseline survey. Of these 14.0% (n = 185) were ineligible as they were aged < 16 years or > 18 years, andfive students provided no data, leaving a baseline sample of 1130 (47.8%) students. Although this equatesto a response rate of < 50%, the number of potentially eligible students at baseline (n = 2363) wasprovided by each institution and overestimates the actual number of new students aged 16–18 years inthat setting, thereby underestimating the true response rate, especially in large FE settings, as a result ofstudents enrolling in principle prior to September but not registering at the start of term, deferring ordropping out in early September; inclusion of students who study across multiple campuses but whoseprimary campus is not the study site; and the inclusion of some students aged > 18 years because ofincomplete information at enrolment. In September 2015, 412 eligible students completed the follow-upsurvey (36.5% of baseline respondents; 17.4% of all potentially eligible students at baseline).

The second objective was to explore the experiences of students, staff and the intervention team. Qualitativedata indicated that implementation was limited by various factors, including staff’s and students’ uncertaintyabout the need for, and appropriateness of, smoking prevention activities in FE settings, the managementof intervention, the high turnover of intervention team staff and the short lead-in time prior to implementation.Although support was expressed for the involvement of external health agencies in the FE setting, the majorityof staff members and students perceived that FE is ‘too late’ for smoking prevention activities, with currentsmokers better served by cessation activities and resistance from non-smokers to educational messages with



xxi

high degrees of familiarity. Significantly, the act of intervention was itself a source of resistance, with both staffand students suggesting that such approaches contrast with institutional cultures in the FE sector aimed atpromoting personal responsibility and developing autonomy in a population transitioning from moreconstrained schools. The emphasis on freedom of choice was expressed via students’ right to smoke.

The third objective was to pilot primary, secondary and intermediate outcome measures and economicmethods. There were low numbers of missing data for all pilot primary and secondary outcomes from thestudent baseline surveys completed in September 2014 (n = 1130 eligible participants) and in the 1-yearfollow-up surveys completed in September 2015 (n = 412 eligible participants). The prevalence of weeklysmoking at baseline was 20.6% and was 17.2% at follow-up. Of the 336 students who were not aweekly smoker at baseline, only 21 (6.3%) reported being a weekly smoker at follow-up. The trial arms werenot well balanced for the indicative primary and secondary outcome measures at baseline or follow-upbecause of the small number of clusters and heterogeneity between clusters (e.g. sixth form and communitycolleges). It was feasible to assess changes in intermediate (process) outcome (e.g. smoking norms/attitudes,self-efficacy, situational resistance skills, etc.) and economic measures (EQ-5D-5L, health service use) overtime. At follow-up, the quantitative process outcomes identified that most students attempting to purchasetobacco were still able to do so. Only 5.1% students were aware of The Filter project at follow-up, althoughthe proportion was higher in the intervention group (7.1%) than in the control group (2.9%).

Conclusion and recommendations

This 1-year pilot study is the first reported evaluation of a universal smoking prevention intervention in anFE context to date, and the first cluster randomised controlled trial (RCT) in FE settings in the UK. It wasnot feasible to implement the Filter FE intervention as planned, and the methods used had low levels ofacceptability among students and staff. FE settings do not appear to be a conducive environment forsmoking prevention intervention activities, although weaknesses in the management of this interventionalso further hindered implementation in this pilot. A larger cluster RCT to examine the effectiveness andcost-effectiveness of this intervention is not recommended. The very low prevalence of smoking uptakesuggests that further consideration is needed on whether prevention or cessation activities would be mosteffective in FE and other educational settings. Findings should be considered in relation to evidence on ageat onset for young smokers. It was feasible to recruit, randomise and retain FE settings within a cluster RCTdesign. FE managers valued the opportunity to be involved in health research and accepted randomisation.However, further methodological work is recommended to improve student recruitment and retentionrates if RCTs are to be conducted in this setting.

Trial registration

This trial is registered as ISRCTN19563136.

Funding

Funding for this study was provided by the Public Health Research programme of the National Institute forHealth Research and by the Big Lottery Fund.

SCIENTIFIC SUMMARY


xxii

Chapter 1 Introduction

Youth smoking: a public health priority

Smoking is a major cause of preventable illness, premature death and health inequalities in the UK.Preventing young people from taking up smoking is vital to maintain and accelerate recent declines insmoking rates. Although much research has been undertaken to develop and evaluate school-basedprevention interventions targeting 11- to 15-year-olds,1 the General Lifestyle Survey Overview: A Report onthe 2010 General Lifestyle Survey (GLS) illustrates that smoking continues to grow rapidly among olderadolescents.2 The GLS does not differentiate those in or out of education; however, with > 1.5 millionBritish 16- to 18-year-olds now enrolled in further education (FE) courses, new smoking preventioninterventions are required that target FE settings (e.g. general FE colleges, ‘sixth form’ colleges attached tosecondary schools, etc.).3 As well as being a period in life when smoking often begins, the transition to FEitself may also increase the risk of smoking as young people are exposed to new sources of peer influenceand have more independence from their parents.

Health improvement in further education settings

Research evidence about smoking prevention interventions delivered in FE settings is sparse. Two recentsystematic reviews of health improvement interventions in educational sites contain no reference to suchstudies in FE settings.4,5 This finding supports calls from the National Institute for Health and Care Excellencefor more evidence regarding smoking prevention interventions in secondary schools and in other youthsettings such as FE institutions.3 Furthermore, the failure of the two reviews4,5 to identify any clusterrandomised controlled trials (RCTs) undertaken within FE settings highlights the lack of rigorous healthimprovement evaluation in this context to date.

A search of bibliographic databases undertaken in 2013 identified a further 14 relevant reports about smokingprevention and other health improvement interventions in FE settings.6–19 Among these, six non-systematicliterature and policy reviews reported increasing policy interest in health improvement interventions targetingyoung people within FE settings, but noted the absence of any evidence regarding appropriate or effectiveinterventions in FE settings.6,7,9,10,12,14 No examples of effective smoking prevention interventions delivered inthis context were identified. Three studies evaluated single-session motivational interviewing interventions inEnglish FE settings,8,11,17 finding that it is feasible to deliver brief interventions within FE settings.11 These studiesalso found that motivational interviewing targeting high-risk students engaged in drug use may reduce theiruse of cigarettes, alcohol and drug use.8 However, it was not an effective method for preventing the uptake ofsmoking among 16- to 19-year-olds in FE.17 One quasi-experimental study of a multicomponent interventioncombining health education, counselling and nicotine therapy in French vocational colleges was found to beeffective in supporting smoking cessation.19

Effective smoking prevention methods and approaches

With no evidence of effective smoking prevention methods or approaches in FE settings, the findings offive recent systematic reviews of smoking prevention interventions delivered in other educational and/orcommunity contexts were identified and synthesised to inform the pilot intervention.20–24 The reviewssuggest that the following smoking prevention methods and approaches are effective: reducing theillicit sale of tobacco products to under-18s;20–23 initiating tobacco-free policies and environmental change;22

age-appropriate, interactive educational messages delivered via intensive, long-term mass media campaigns;21

and social competency and skills development interventions to support young people to resist peer influence.24



1

A recent systematic review of school effects/environment interventions also found that initiating tobacco-freepolicies and environmental change can be effective, especially in permissive contexts,4 which is likely to be thecase in some FE settings.

This evidence highlights the relevance of multilevel smoking prevention interventions and identifies a set ofintervention methods and approaches that may underpin intervention efficacy:

l restricting the availability of tobacco and opportunities for smokingl restructuring environmental contextsl educating and persuading young people about the harms of smoking and social norms via multiple

methods and communication channelsl modelling social/situational resistance skills.

Systematic reviews also consistently find that ‘multilevel’ interventions, which address both individual andenvironmental determinants of behaviour simultaneously, are most effective for improving young people’shealth outcomes.20,23,25,26 These interventions, which include ‘higher-level’ environmental components, alsotend to be more cost-effective,27 and are less likely to generate inequalities than individually focusedcomponents alone.28,29 However, if such interventions are to deliver major public health gains, they mustalso be feasible to deliver and sustain.30

‘The Filter FE’ intervention design and logic model

‘The Filter FE’ intervention was co-designed by Action on Smoking and Health (ASH) Wales and theresearch team following a commissioned call from the National Institute for Health Research (NIHR) PublicHealth Research (PHR) programme in 2013. It is a smoking prevention intervention managed and deliveredby trained staff working on ASH Wales’ ‘The Filter’ youth project, who apply existing staff training, socialmedia and youth work resources in FE settings. Informed by the socioecological theory of health,31 andevidence of effective smoking prevention methods and approaches (summarised in Effective smokingprevention methods and approaches), Filter FE aimed to integrate multiple intervention activities within amulticomponent, multilevel intervention for FE settings.

The intervention design and hypothesised mechanisms are summarised in the logic model (Figure 1) anddescribed in more detail in Chapter 2, Intervention components. In summary, five areas of synergisticactivity were planned to augment any existing activities already undertaken in FE settings: (1) working withlocal shops to restrict the sale of tobacco to under-18s; (2) implementing tobacco-free campus policies;(3) training FE staff to deliver smoke-free messages; (4) publicising The Filter youth project’s online campaigns,advice and support services via FE websites and social media; and (5) on-site youth work activities to providecredible educational messages and promote social/situational resistance skills, as well as signposting cessationservices. As described in the logic model (see Figure 1), it was hypothesised that these components wouldprevent the uptake of smoking via the restriction of the availability of tobacco; restructuring the institutionalcontext to prevent smoking on site and promote non-smoking behaviour as normative; education andpersuasion of young people regarding the harms of smoking and social norms via multiple interactivemethods and channels of communications; and modelling social/situational self-efficacy and resistance skills.

In order to enable scalability across different types of FE settings (including large institutions), as well assustainability and fidelity, the intervention was designed so that it involved standardised processes andactivities balanced with opportunities for a degree of local tailoring of activities. Some flexibility to allow forlocal adaptation can support universal adoption, institutional ownership and sustainable implementationof multiple activities.32,33 The intervention was also designed to allow the ‘dose’ of staff training and youthwork activities to vary according to the size of institutions.

INTRODUCTION


2

Intervention staffing:standard inputs

Implementation processes to ensure the activities

meet local institutional needs

Levels and determinantsof behaviour change

(pilot intermediate outcomes)

Hypothesised health impacts (pilot primary and

secondary outcomes)

Mapping of local retailersand liaison with local

authority tobacco action group and trading standards

Community level

Restricting the availability of tobacco (student survey

and mystery shopper audit)

Institutional level

Tobacco-free environment(FE college policy audit

and structured observations)

Staff commitment to smoking prevention and

confidence in delivering key messages (staff and

student surveys)

Individual level

Increased awareness of The Filter social media and

support services(student survey items)

Increased knowledge about harms and social norms

(NatCen knowledge/belief items)

Improved social/situational self-efficacy and skills

(ESFA items)

Letters and visits to local shops to inform retailers of the project and trading standards penalties

The interventionmanager(s)

Management and co-ordination of all other staff

The Filter web and social media officers

Support FE managers to review and revise local

institutional policy

Application of‘tobacco-free campus’ guidance

to improve college policies

The Filter training andeducation officers

Staff briefings, audit of current practice and

planning of staff training

Meetings with IT staff and student groups to map

out and review social media

Staff support tobacco-free environment policies

Training in antismokingmessages and skill development

(two, four or six × 2-hour sessions)

Integration of The Filter social marketing campaigns and services

(e.g. text support) with college social media/web

Reinforces youth work messages on risk and norms

The Filter youth development officers

Work with college staff to organise new youth work

activities focussed on antismoking messages,

norms and social competency development

Signpost social media and support services

Youth work activities delivered on-site

(one graffiti wall/art day; five, 10or 15 × 2-hour group work sessions;

‘Cut Films’ promotional event)

Primary outcome

Incidence of smoking weekly or more

(ONS GLS adapted)

Secondary outcomes(all students)

Prevalence ever smoked (ONS GLS)

Frequency of cannabis use in last 30 days

(EMCDDA EMQ)

High-risk alcohol use reported (AUDIT-C)

Health-related QoL (EQ-5D-5L)

Secondary outcomes (baseline smokers)

Cigarettes per week among non-quitters

(ONS GLS)

Core intervention activities (’dose’ proportionate to

institutional size when applicable)

Cessation in the last year (ONS GLS)

Prevalence of nicotine dependence among non-quitters

(HSI measure)

FIGURE 1 Intervention logic model. AUDIT-C, Alcohol Use Disorders Identification Test Consumption; EMCDDA, European Monitoring Centre for Drugs and Drug Addiction;EMQ, European Model Questionnaire; EQ-5D-5L, EuroQol-5 Dimensions, 5-level version; ESFA, European Smoking Prevention Framework Approach; HSI, Heaviness of SmokingIndex; IT, information technology; ONS, Office for National Statistics; QoL, quality of life.

DOI:10.3310/phr05080

PUBLIC

HEA

LTHRESEA

RCH2017

VOL.5

NO.8

©Queen

’sPrinter

andController

ofHMSO

2017.This

work

was

producedby

Fletcheret

al.under

theterm

sof

acom

missioning

contractissued

bythe

Secretaryof

Statefor

Health.

Thisissue

may

befreely

reproducedfor

thepurposes

ofprivate

researchand

studyand

extracts(or

indeed,the

fullreport)may

beincluded

inprofessionaljournals

providedthat

suitableacknow

ledgement

ismade

andthe

reproductionisnot

associatedwith

anyform

ofadvertising.

Applications

forcom

mercialreproduction

shouldbe

addressedto:

NIHRJournals

Library,NationalInstitute

forHealth

Research,Evaluation,

Trialsand

StudiesCoordinating

Centre,

Alpha

House,

University

ofSoutham

ptonScience

Park,Southam

ptonSO

167N

S,UK.

3

Public involvement

As well as co-designing the pilot intervention, staff working on ASH Wales’ The Filter youth project wereinvolved in designing all aspects of the pilot trial and process evaluation prior to bid submission. The researchteam also worked with the Involving Young People Officer based in the Centre for the Development andEvaluation of Complex Interventions for Public Health Improvement to organise two consultations with theAdvice Leading to Public Health Advancement (ALPHA) youth group at the project development stage. ALPHAis a group of young people aged 14–21 years who advise researchers on intervention design, logic modellingand data collection methods by discussing and debating their views on the proposed research. Staff fromPublic Health Wales and FE teachers were also consulted on the intervention design, logic model andresearch strategy.

Three further consultation meetings with the ALPHA group took place post commissioning to enablethe researchers to consult with young people during the project on recruitment and survey methods(e.g. advice on the design of publicity materials, information sheets and e-questionnaires), strategies forincreasing retention/follow-up, and public engagement and knowledge exchange activities.

Study aim, objectives and research questions

The aim of the pilot trial was to evaluate the feasibility and acceptability of implementing and trialling anew multilevel smoking prevention intervention in FE settings. The study had three objectives.

The first objective was to assess whether or not prespecified feasibility and acceptability criteria were met,which were agreed with the NIHR Evaluation, Trials and Studies Coordinating Centre and Trial SteeringCommittee, and deemed necessary conditions for progressing to a Phase III trial. The progression criteriaare listed in full in Chapter 2, Progression criteria. In order to meet this objective, data were collected andanalysed to address the following research questions (RQs):

1. Did the intervention activities occur as planned in (at least) two out of three intervention settings?2. Were the intervention activities delivered with high fidelity across all settings?3. Was the intervention acceptable to the majority of FE managers, staff, students and the intervention

delivery team?4. Was randomisation acceptable to FE managers?5. Did (at least) two out of three colleges from each of the intervention and control arms continue to

participate in the study at the 1-year follow-up?6. Do student survey response rates suggest that we could recruit and retain at least 70% of new

students in both arms in a subsequent effectiveness trial?

The second objective was to explore the experiences of FE students, staff and the intervention deliveryteam to refine the intervention and study design prior to a potential Phase III trial. In order to meet thisobjective, data were collected and analysed to address the following RQs:

7. What are students’, college staff’s and intervention team members’ experiences of the interventionand views about its potential impacts on health?

8. What are the barriers to, and facilitators of, implementation and how do these vary according tocollege context and/or other factors?

9. Were there any unexpected consequences?10. How acceptable were the data collection methods to students and staff, and do participants think that

longer-term follow-up via e-mail or telephone interview would be feasible?11. What resources and partnerships are necessary for a Phase III trial?

INTRODUCTION


4

The third objective was to pilot primary, secondary and intermediate outcome measures and economicevaluation methods prior to a potential Phase III trial. It was not an objective of the pilot study to assessintervention effects nor was it powered to do so, but data were collected and analysed to address thefollowing RQs:

12. Does the primary outcome measure (smoking weekly or more) have an acceptable completion rate,adequate validity and minimise floor/ceiling effects?

13. Do cotinine concentrations of saliva samples indicate any evidence of response bias between arms inself-reported smoking status?

14. Was it feasible and acceptable to measure all the secondary and intermediate outcomes of interest atbaseline and follow-up?




5

Chapter 2 Methods

In this section of the report we provide an overview of the study design, including the specificintervention components examined. Details are then provided of the sampling and recruitment of the FE

settings and randomisation. We then describe the methods used to assess the ‘progression criteria’(objective 1), explore participants’ experiences of the process of implementing and trialling the intervention(objective 2) and examine pilot trial outcomes (objective 3). Details of the pilot economic analysis and trialregistration, governance and ethics are provided at the end of this chapter.

Study design: overview

A cluster randomised controlled pilot trial was undertaken in six FE settings in south-east Wales withallocation to the Filter FE intervention (three settings) or continuation of normal practice (three settings).In order to assess the feasibility and acceptability of delivering and trialling the intervention according toprespecified criteria (objective 1), we collected a range of quantitative and qualitative data via semistructuredobservations of the intervention delivery, interviews with FE college managers and the intervention team,and documentary evidence (e.g. college policies, intervention team records, etc.). The retention of FE settingsand response rates were assessed using student survey data.

To explore participants’ experiences of implementing and trialling the Filter FE intervention (objective 2),data were collected via semistructured interviews with FE managers and the intervention team, focusgroups with students and staff, as well as additional process and contextual data via observations ofintervention settings, staff training and youth work activities.

Primary, secondary and intermediate (process) outcomes, and economic evaluation methods were alsopiloted in this study (objective 3). Surveys of new students enrolling at the participating FE settings inSeptember 2014 (baseline) and September 2015 (1-year follow-up) were used to examine the pilotprimary, secondary and economic outcome measures. Informed by the intervention logic model, multiplesources of data were also collected at baseline and follow-up to pilot intermediate (process) outcomes atmultiple levels: the restriction of the availability of tobacco in local shops was assessed via ‘mysteryshopper’ audits; changes to the institutional environment and policies were assessed via structuredobservations and analysis of college policy documents; and students’ knowledge, norms and social/situational self-efficacy and resistance skills were assessed via the student survey.

Intervention componentsThis section describes how each of the pilot intervention components were intended to be delivered, bywhom, and their logic.

Prevention of the sale of tobacco to further education students aged < 18 yearsTo restrict availability locally, the intervention manager would map and contact all shops selling tobaccowithin 1 km of the intervention setting (i.e. within a 10-minute walk). Information letters would bedistributed to these retailers to inform them that a new project (Filter FE) was taking place at their local FEinstitution, explain why reducing supply is an important component of prevention and remind them aboutthe penalties for selling tobacco to under-18s. The letter focused only on sales of legal tobacco throughthe retailers. Posters, stickers and other materials would also be supplied for these shops to provideinformation to their customers about the legal age for purchasing tobacco products and the requirementsto produce statutory identification to purchase tobacco.

Institutional policy review to promote a tobacco-free environmentTo restrict opportunities for smoking and promote non-smoking as the norm via modifying the institutionalcontext, the intervention manager would work with FE managers to review institutional policies using the



7

tobacco-free campus guidance developed by ASH Australia.34 This tool uses a three-stage process topromote a tobacco-free environment, including advice on advertising, the supply of tobacco and supportservices, as well as information on maintaining smoke-free public areas, buildings and vehicles. First,current policies and practices are reviewed using this tool to develop a new whole-campus tobacco-freepolicy. Second, the revised policies are implemented and launched. Third, policies are monitored, evaluatedand updated/refined if required.

Further education staff trainingTo train staff to deliver smoke-free educational messages and support institutional change, training officersemployed on The Filter youth project [accredited by the Young Men's Christian Association (YMCA) andAgored Cymru] would organise and deliver training sessions on site using modules and teaching resourcesdeveloped and piloted by ASH Wales in schools and other youth settings. Interactive, 2-hour trainingworkshops would be delivered to approximately 10 staff per session, with FE staff trained to integrateactivities about smoking into their lesson plans and other routine work (e.g. via body mapping the healthharms of smoking, exercises on how tobacco companies recruit young smokers). All staff attending thesesessions would also be encouraged to champion new tobacco-free policies (see Institutional policy review topromote a tobacco-free environment) and intervene to prevent smoking on site. The number of sessionsto be delivered would vary depending on the size of the FE setting to ensure that resources are distributedappropriately: one session to be delivered at smaller ‘sixth form’ sites (i.e. to reach a total of approximately10 members of staff) and 2–4 sessions to be delivered at medium and large FE campuses, respectively(to reach up to 20–40 members of staff).

Social mediaTo educate and persuade students about the harms of smoking, social norms and the relevance of supportservices, The Filter youth project’s web and social media officers would work with staff and students tointegrate it online social marketing campaigns, advice and support services (e.g. The Filter text/instantmessaging services) with institutional websites and social media channels maintained by staff and/or students[e.g. the college Facebook (www.facebook.com; Facebook, Inc., Menlo Park, CA, USA) page, institutionalTwitter (www.twitter.com; Twitter, Inc., San Francisco, CA, USA) feeds, Instagram (www.instagram.com;Facebook, Inc., Menlo Park, CA, USA), etc.]. As well as embedding information on each intervention setting’shome/index webpage, the web and social media officers would work with the college information technologystaff and consult students to identify opportunities for publicising key information and messages via frequentlyaccessed webpages/micro-sites (e.g. online learning portals, e-mail login pages).

Youth work activitiesTo educate and persuade students about the harms of smoking and model social/situational resistanceskills, qualified youth workers from The Filter project would work with college staff and students to planand deliver a range of youth work activities on site (e.g. smoke-free message film-making, graffiti wallsand/or other arts-based activities). Youth workers would launch the project in the autumn term, and thenwork with staff and/or student groups to identify 5, 10 or 15 groups (depending on institutional size) of10–20 students to take part in locally tailored group-based activities. As with the staff training, the numbersof sessions delivered would vary according to the FE setting’s size to ensure resources are distributedappropriately. These group-based youth work activities would be provided on site during college time andtypically last 1–2 hours. Students would not be targeted based on their smoking status or any othercharacteristics, as the aim is to recruit as many newly enrolled students as possible. Information aboutonline support/advice services would also be provided to current smokers, when appropriate.

Sampling and recruitment of further education settings

The following diversity and matching criteria were used to purposively sample six FE settings in south-eastWales: large FE college campuses (new intake of > 500 students) (n = 2), small FE college campuses(new intake of < 500 students) (n = 2) and ‘sixth form’ colleges attached to schools (n = 2).

METHODS


8

http://www.facebook.com

http://www.twitter.com

http://www.instagram.com

Purposive sampling ensures a diversity in contexts at the intervention piloting stage so that a realist lenscan be applied to address questions regarding not only what is feasible and acceptable in general, butalso for whom and under what circumstances, and place much more emphasis on exploring potentialmechanisms of action and how these may vary by context prior to large-scale RCTs.35 To avoid contexts inwhich implementation may be less challenging (or atypical in other ways), private institutions, small sites(with < 100 students) and ‘sixth forms’ at schools where < 10% of students are entitled to free schoolmeals were not included. To minimise the potential for contamination across arms, no more than oneFE setting was recruited from any middle-layer super output area, nor were FE settings recruited inneighbouring middle-layer super output areas.

The intervention team identified and contacted school and FE managers in the summer term of 2014, withrecruitment complete by July 2014. A total of 10 FE settings in south and mid-Wales were contacted byASH Wales’ staff in May and June 2014. Those who first expressed an interest were visited by researchersbetween June and July 2014, until six FE settings had been recruited according to the sampling criteria,above. Participating FE settings are listed in Table 1 according to recruitment strata (pseudonyms).

Randomisation

Colleges agreed to take part in the study prior to randomisation. This study used a 1 : 1 allocation ratio.Allocation to intervention and control arms was conducted by the study statistician and stratified by thesize and type of FE settings. The three strata were large FE college campuses with a new intake of > 500students, small FE college campuses with a new intake of < 500 students and sixth forms within secondaryschools. Table 1 reports the outcome of the random allocation to trial arm by sampling strata, includingthe size of each FE setting’s new intake.

It was not possible for colleges, the intervention team and researchers to be blinded to allocation throughoutthe study. However, colleges were randomised after baseline data collection to ensure that all students,staff and researchers were blind at the time of the recruitment of colleges and baseline data collection.

Progression criteria

In line with Medical Research Council guidance,36 data collection during the pilot trial focused on assessingacceptability and feasibility, and allowing us to judge progress against the agreed criteria for progressionto a subsequent trial of effectiveness and cost-effectiveness. The first objective was specifically to assesswhether or not the criteria deemed necessary in order to progress to a larger cluster RCT were met (Box 1).These were agreed by the investigator team, the NIHR Evaluation, Trial and Studies Coordinating Centreand Trial Steering Committee prior to commencing the pilot trial as evidence of feasibility and acceptabilityof the intervention and trial methodology.

TABLE 1 Participating FE settings and allocation to trial arm

Sampling/randomisation strata FE setting (pseudonyms) AllocationEstimated new studentsaged 16–18 years

Large FE college campuses (n= 2) Valeside College Intervention 1027

Middledale College Control 760

Small FE college campuses (n = 2) Laurelton College Intervention 130

Glynbel College Control 175

School ‘sixth form’ colleges (n= 2) Athervale Sixth Form Intervention 110

Afonwood Sixth Form Control 161



9

Data sourcesIn order to answer RQ 1 (see Box 1), quantitative and qualitative data were collected from multiplesources. To assess if tobacco retailers within 1 km of the FE setting were contacted in writing within3 months of the start of the intervention, data collected via intervention team records were examined andcross-checked through interviews with the intervention team. To assess if institutional policies and practiceswere reviewed and updated using tobacco-free campus guidance, with changes communicated to staffand students within 6 months of the start of the intervention, data collected via intervention teamchecklists were examined and cross-checked with documentary analyses of college policies and structuredobservations of the FE environments at follow-up. To examine the fidelity and reach of staff training, datacollected via intervention team checklists and semistructured observations of training were examined.Integration of ASH Wales’ The Filter online resources by intervention setting was assessed usingintervention team checklists and cross-checked via semistructured observations of college websites andsocial media channels. To examine the implementation of youth work activities, data collected viaintervention team checklists were examined and cross-checked in interviews with FE managers.

In order to answer RQ 2 (see Box 1), semistructured observations of staff training sessions (n = 1 perintervention setting) and group-based youth work sessions (n = 1 per intervention setting) were used toassess fidelity of delivery of those components across settings and the fidelity of other interventioncomponents (activities aiming to prevent the sale of tobacco to under-18s in shops near the interventionsite, institutional policy review and revision, social media integration) were examined via intervention teamchecklists and interviews with the intervention team and FE managers. To answer RQ 3 (see Box 1),intervention acceptability and whether or not this was reported by the majority of participants, wasassessed via data from semistructured interviews with FE managers and the intervention team, and student

BOX 1 Research questions addressing the agreed criteria for progression

1. Did the intervention activities occur as planned in (at least) two out of three intervention settings? This will

be assessed according to the extent to which the following intervention activities occurred:

¢ tobacco retailers within 1 km of the FE setting were contacted in writing within 3 months of the start of

the intervention¢ institutional policies and practices were reviewed, updated using the tobacco-free campus guidance,

and changes communicated to staff and students within 6 months of the start of the intervention¢ a minimum of one, two or four staff training sessions were delivered as planned (according to

institutional size), with a minimum of five staff attending each session¢ The Filter youth project’s web-based information, advice and support services were embedded on the FE

institution’s homepage during the intervention and online information, advice and support services are

promoted through at least one local social media channel maintained by staff and/or students (e.g. the

college Facebook page, Twitter feed, etc.)¢ a minimum of 5, 10 or 15 youth work sessions were delivered as planned (according to institutional

size) with a minimum of eight different students attending each session.

2. Were the intervention activities delivered with high fidelity across all settings?

3. Was the intervention acceptable to the majority of FE managers, staff, students and the intervention

delivery team?

4. Was randomisation acceptable to FE managers?

5. Did (at least) two out of three colleges from each of the intervention and control arms continue to

participate in the study at the 1-year follow-up?

6. Do student survey response rates suggest that we could recruit and retain at least 70% of new students in

both arms in a subsequent effectiveness trial?

METHODS


10

and staff focus groups. Data from semistructured interviews with FE managers were used to examine RQ 4(see Box 1). In order to answer RQs 5 and 6 (see Box 1), the retention of FE settings and response rateswere assessed using student survey data.

Data analysis methodsThe quantitative and qualitative data collected to answer RQs 1–6 were managed and analysed separately.For example, the standardised checklist data from the mystery shopper, environmental observation andpolicy audits were collated and analysed in Microsoft Excel® 2013 (Microsoft Corporation, Redmond, WA,USA), whereas data from focus groups and interviews were analysed using NVivo version 10 (QSRInternational, Warrington, UK). The emergent results from each data source were shared and discussedamong the research team. Individual results were collated according to a framework derived from key RQsand relevant progression criteria to ensure that data from all sources were used when pertinent to answerthe RQs, and to facilitate triangulation. When data from one source contradicted data from another source,this was noted and discussed.

Evaluating participants’ experiences of the process

In addition to examining intervention delivery according to prespecified criteria (objective 1), a secondobjective was to explore student, staff and intervention team experiences of implementing and trialling theintervention, and how this varied in different FE contexts, in order to refine the intervention and trialmethods. RQs 7–11 addressed this objective (Box 2).

To answer RQs 7–11, multiple sources of qualitative data collected via semistructured observations,semistructured interviews and focus groups were analysed to explore student, staff and interventiondelivery team experiences in depth, and how and why these varied.

Qualitative process dataSemistructured interviews and focus groups were conducted with a range of stakeholders to explore indetail the process of planning, implementing and receiving the intervention. Table 2 summarises theprocess evaluation data collected at each FE site. Each method of data collection is described in moredetail below.

Semistructured observationsSemistructured observations of staff training and youth work sessions were conducted to providecontextual detail on the delivery, and receipt, of the intervention and to provide data on potential barriersto, and facilitators of, implementation. Observations focused on the way in which sessions were delivered,the content and activities included, and the way in which they were received by staff and young people.Semistructured observations (of staff training and youth work sessions) were recorded on templates

BOX 2 Research questions to evaluate participant experiences

7. What are students’, college staff’s and intervention team members’ experiences of the intervention and

views about its potential impacts on health?

8. What are the barriers to, and facilitators of, implementation and how do these vary according to college

context and/or other factors?

9. Were there any unexpected consequences?

10. How acceptable were the data collection methods to students and staff, and do participants think that

longer-term follow-up via e-mail or telephone interview would be feasible?

11. What resources and partnerships are necessary for a Phase III trial?



11

devised a priori, documenting the content of sessions, the number and types of participants, and thedynamics among the group, allowing for observation of intervention acceptability and fidelity.

Whenever possible, the same researcher conducted the observations, interviews and focus groups in eachFE setting, and observations of staff training and youth work sessions were carried out before focus groupsto maximise the opportunity to respond to issues that arose from observations and preliminary analysesin focus groups and interviews. This depth of fieldworker immersion and data triangulation enabledunexpected and emergent issues [e.g. attitudes and beliefs around electronic cigarettes (e-cigarettes)] tobe tested and clarified.

One staff training session in each of the three intervention sites was observed between April and June 2015.Sessions lasted a maximum of 2 hours. A total of 28 members of staff attended the three sessions observed(Table 3). Participants included both teaching and support staff.

Three observations of youth work sessions were undertaken, one in each intervention site, between Marchand May 2015. The largest observed group had 22 students, the smallest had six participants; a memberof teaching staff sat in on each of the observed sessions (Table 4). It is not known what proportion ofparticipants identified as smokers, as it was not the intention to target students based on smoking status.

Focus groups with students and staffFocus groups were conducted with students and staff at all the intervention settings to explore their views onstudent smoking norms and behaviour; their awareness and/or experiences of participation in Filter FE projectactivities at their college, including their views on how successfully each component was implemented andbarriers to implementation; their perceived impact of the intervention activities on student and staff smokingnorms, and behaviours at their college; and the acceptability and feasibility of recruiting and collectingmultiple waves of e-survey data from students. Focus groups with students were chosen for both pragmatic

TABLE 2 Qualitative process data collected by arm and setting

Trial arm and FE setting

Data collection method (n)

Semistructured observations of

Studentfocusgroups

Stafffocusgroups

FE manager/staffinterviews

Staff trainingsessions(total delivered)

Youth worksessions(total delivered)

Intervention arm

Valeside College(large FE college campus)

1 (1) 1 (10) 5 2 1

Laurelton College(small FE college campus)

1 (1) 1 (2) 4 2 1

Athervale Sixth Form(school sixth form college)

1 (1) 1 (3) 2 1 1

Comparison arm

Middledale College(large FE college campus)

N/A (0) N/A (0) N/A N/A 1

Glynbel College(small FE college campus)

N/A (0) N/A (0) N/A N/A 1

Afonwood sixth form(school sixth form college)

N/A (0) N/A (0) N/A N/A 0

N/A, not applicable.

METHODS


12

and methodological reasons. First, focus groups allowed us to quickly capture a range of views from arelatively largely number of students (n = 69 in all focus groups). Second, we wished to gain insights intostudents’ shared (or contested) understandings of the smoking culture within FE settings; focus groups were,therefore, the most appropriate method.

The number of student focus groups varied according to the size of the college (see Table 2). The aim wasto recruit six student focus groups (three smokers groups, three non-smokers groups) at the large FEcollege campus, four in the medium-sized college campus and two in the sixth form college. This wasalmost achieved: in the large FE college, five student focus groups were taken, with four and twoundertaken as planned in the other intervention colleges. FE managers recruited students to attend thefocus groups. They were asked to recruit both students who identified as smokers and those whoidentified non-smokers, from a range of courses. In practice, it was difficult to purposively sample andstratify students into groups by smoking status as some young people identified as non-smokers butrevealed that they smoke (e.g. ‘social smoking’) during the focus group; other groups were mixed becausethe students were recruited through friendship groups or through a convenience sample of one tutorgroup. Although managers were briefed about how to recruit students, this mode of recruitment meantthat there was potential for students to come along without a clear understanding of the purpose of thefocus groups. We were therefore careful to provide a clear introduction before each focus group started,allowing participants to withdraw if they wished.

Student focus groups took place between April and June 2015. Most were in June 2016, and wereconducted only after all intervention activities were completed at the site. A total of 69 studentsparticipated in the focus groups, of whom 31 were female (45%) and 18 explicitly identified themselves assmokers (26%); some additional participants identified as non-smokers but during the focus groupdiscussion revealed they sometimes smoked. The smallest focus group had two students and the largesthad 13 participants. The focus groups included part-time and full-time students from a range of courses,including Advanced level (A-level), Business and Technology Education Council (BTEC) and vocationalstudents. Student focus groups lasted between 40 and 80 minutes, and were all conducted in privaterooms at the intervention college sites using topic guides. Topic guides covered their views on studentsmoking norms and behaviours; awareness of and participation in The Filter FE project; how successfullyeach component was implemented and why implementation may have been limited; perceived impact on

TABLE 3 Participants in semistructured observations of staff training sessions

Intervention setting

Participants (n)

Total Teaching staff Support staff

Valeside College (large FE college campus) 6 0 6

Laurelton college (small FE college campus) 9 0 9

Athervale Sixth Form (school sixth form college) 13 10 3

TABLE 4 Participants in semistructured observations of youth work sessions

Intervention setting

Participants (n)

Teachers present (n)Total Female Male

Valeside College (large FE college campus) 8 4 4 1

Laurelton College (small FE college campus) 6 0 6 1

Athervale Sixth Form (school sixth form college) 22 16 6 1



13

student and staff smoking norms behaviours; and the acceptability and feasibility of recruiting and datacollection methods piloted. All participants were provided with a £10 ‘Love2shop’ voucher for taking partin the focus groups.

We conducted two staff focus groups at the two larger intervention sites: one focus group with staff whohad received training from The Filter team as part of the intervention, one with staff who had not. Onlyone staff focus group took place at Athervale Sixth Form because of the smaller number of students andstaff based there. We aimed to include approximately eight staff in each focus group but, in practice, itwas difficult to recruit staff to attend the focus groups, especially those who had not attended thetraining. In total, 19 staff participated across the five focus groups (including eight staff trained by TheFilter team); the size of focus groups ranged from two to six participants. Focus groups were conducted inprivate rooms on each site using semistructured topic guides that covered the same broad areas as the oneused in the student focus group (see above paragraph). Participating staff represented a range of teaching,management and support positions within each setting.

Semistructured interviews with further education managersSemistructured interviews were conducted with FE managers at both intervention and control sites toexplore their experiences of participating in a trial, including the acceptability of randomisation and datacollection methods; perceived benefits and challenges of the Filter FE intervention on student and/or staffsmoking behaviours; and, at intervention sites, managers’ experiences of implementation in their collegecontext, including activities completed/not completed and barriers to, and facilitators of, implementation.Interviews were used as FE managers were the person (or in the case of one paired-interview, people) whocould offer the best insight into their experience of participating in the research.

A member of the management staff at each of the six participating FE settings was recruited to participatein a semistructured interview at the end of the intervention. Interviews were conducted with FE managersat both intervention and control sites between June 2015 and February 2016, after the completion of theintervention. One (control) FE manager declined to participate in the interview. One interview was face toface with two staff members who had been working jointly as the lead FE manager for the study and fourinterviews were conducted over the telephone at the participants’ request. This may have affected the datacollected as the face-to-face interview was longer and the interviewees and interviewer had an establishedrapport. However, the telephone interviews were conducted by two experienced qualitative researchers(one was the researcher who conducted the face-to-face interview) and we ensured that all data wereincluded in the analysis and contributed to the findings. We were careful to provide prompts to aid therecall of managers who were interviewed. The time between recruitment to the study and the interviewsmay have affected some interviewees’ accounts and led to difficulties in recall, although two of theinterviewees were particularly keen to provide feedback about the recruitment process and the early stagesof the study, suggesting that the interviewees had time to reflect on the process even if they could notremember fine details. Interviews lasted between 20 and 60 minutes. Topic guides included questions andprompts regarding our a priori progression criteria, their experiences of being involved in the pilot RCT andof planning, implementing and receiving the intervention (if they were an intervention site).

Semistructured interviews with the intervention teamSemistructured interviews were conducted with the intervention team to explore their experiences ofimplementing the intervention; facilitators of, and barriers to, implementation; potential changes to bemade to the intervention; and their experience of delivering the Filter FE within a trial context. Two ofthese interviews were conducted before the end of the intervention because the members of staff wereleaving the organisation. Recruitment to these interviews was based on whether or not the individual hadbeen involved in intervention implementation and aimed to encompass a range of different staff roles(project managers, staff trainers, youth workers and social media team members); researchers recruitedthese staff directly.

METHODS


14

Interviews were conducted with six members of the intervention delivery team between April andSeptember 2015. All interviews were conducted face to face; five of the six interviews were conducted inprivate rooms in the ASH Wales offices and one was conducted in a private space at an intervention siteafter a staff training session. Interviews lasted between 30 and 60 minutes. The interview topic guidescovered their experiences of implementing The Filter in FE settings; facilitators of, and barriers to,implementation; what they would change about the intervention if they could; and their experience ofdelivering the intervention within a trial context.

Data analysis methodsQualitative data collected via focus groups and semistructured interviews were transcribed verbatim andentered into NVivo software to aid data management and analysis. Qualitative data were analysed usingtechniques associated with thematic content analysis and grounded theory. Transcripts were initiallydivided into four groups to assist manageability: student focus groups, staff focus groups, FE managerinterviews and intervention team members. First, one of the research team (MW) read each of the groupsof transcripts to familiarise herself with the data. All transcripts for each group were then reread andcoded line by line to identify emergent themes, and an initial coding framework was developed to identifykey themes and subthemes. The coding framework included both deductive codes, derived from key RQsand relevant progression criteria, and inductive codes, identifying other relevant themes emerging from thedata (e.g. e-cigarette use in FE settings). Micky Willmott and a second researcher (RL) independentlyapplied this coding framework to three transcripts, then met to discuss and further refine the codingframework. The final agreed coding framework was applied to all subsequent manuscripts by MickyWillmott, noting and discussing any substantial additions or modifications with the research team, asnecessary. The way in which themes inter-related and how they varied between different groups andcontexts was carefully scrutinised throughout the analysis process and recorded using detailed memos.

The quotations presented in the report were selected because they best illustrate the common and/orinteresting ideas and themes emerging from the data. These were discussed and agreed among theresearch team. When contradictory data were identified (e.g. the difference between school sixth formsand colleges, or when some students identified that they felt that they had been bullied into smoking),these are noted in the report.

Observation records of staff training and youth work sessions were analysed separately from interview andfocus group data by Micky Willmott, with NVivo used to support cross-checking and data triangulation.Records were coded line by line, then grouped according to whether they related to the frameworkdescribed above or inductive, emergent themes.

Most of the qualitative data collected for the process evaluation were collected by experienced qualitativeresearchers who were independent of the trial management or baseline data collection. Although thisadded an additional layer of contacts for FE managers to liaise with, it meant that the researchers hadminimal knowledge or preconceptions about the sites. The researchers were all white professionals (threefemale and one male), which reflected the predominant ethnicity in FE sites. They had little knowledge ofthe areas in which the FE sites are based, and all were English, not Welsh or Welsh speakers, which mayhave had an impact on how the students and staff responded to them in focus groups. Researchers sharedtheir reflections on data collection to aid analysis and an interpretative approach.

Pilot outcome measures

The final objective was to pilot primary, secondary and intermediate outcome measures and economicevaluation methods. The pilot cluster RCT design enabled a range of outcome measures and datacollection methods (student surveys, policy audits, environmental observations and mystery shopper visits)to be piloted at baseline (September 2014) and at the 1-year follow-up (September 2015) to answer RQs12–15 (Box 3).



15

The pilot outcomes are described first in Pilot primary and secondary outcome measures and Pilotintermediate outcome measures, followed by a description of the data sources used to operationalise thesemeasures to assess change at the student, college and community levels (see Data sources). These pilotoutcomes are also summarised in the intervention logic model (see Figure 1). Finally, the methods foranalysing the student survey data and other quantitative data sources are described in Statistical analyses.

Pilot primary and secondary outcome measuresAll the pilot primary and secondary outcome measure data were collected via student surveys at baselineand at the 1-year follow-up, which are described in more detail later (see Data sources).

The pilot primary outcome was prevalence of weekly smoking (defined as smoking at least one cigaretteweekly or more) at the 1-year follow-up, which was assessed using an item adapted from the GLS.2

Students were asked: ‘Do you smoke cigarettes at all nowadays?’ and given four response options: ‘Yes,every day’, ‘Yes, at least once a week’, ‘Yes, occasionally but less than once a week’ and ‘No, never’.Those who responded ‘Yes, every day’ or ‘Yes, at least once a week’ were considered weekly smokers.

The pilot secondary outcomes were lifetime smoking, using the Office for National Statistics (ONS) GLS item;2

use of cannabis in the past 30 days and frequent cannabis use (four or more times in the past 30 days), usingthe European Monitoring Centre for Drugs and Drug Addiction European Model Questionnaire items;37

high-risk alcohol use, using the Alcohol Use Disorders Identification Test Consumption (AUDIT-C) measure;38

and health-related quality of life, using the EuroQol-5 Dimensions, 5-level version (EQ-5D-5L) measure.39

The following three measures were additional pilot secondary outcomes at follow-up for baseline smokersonly: self-report smoking cessation, using the ONS GLS item;2 number of cigarettes smoked per week,using the ONS GLS item;2 and nicotine dependence as measured using the Heaviness of SmokingIndex (HSI).40

Pilot intermediate outcome measuresAs illustrated in the intervention logic model (see Figure 1), it was hypothesised that the interventioncomponents would prevent the uptake of smoking through triggering changes at the individual (student),college and community levels. For this reason, intermediate outcome variables were piloted at each ofthese three levels by collecting a range of additional quantitative process data at baseline and at the 1-yearfollow-up via college policy audits, environmental observations and mystery shopper visits, as well as viathe student surveys at each site.

At the individual level, baseline and follow-up student surveys assessed self-reported changes over timeto attitudinal and knowledge-based precursors to smoking, including perceived prevalence of smoking(i.e. perceived norms), by adapting NatCen items;41 social and situational self-efficacy and skills, using theEuropean Smoking Prevention Framework Approach (ESFA) items;42,43 and awareness of The Filter project.

BOX 3 Research questions to evaluate outcome measures

12. Does the primary outcome measure (smoking weekly or more) have an acceptable completion rate,

adequate validity and minimise floor/ceiling effects?

13. Do cotinine concentrations of saliva samples indicate any evidence of response bias between arms in

self-reported smoking status?

14. Was it feasible and acceptable to measure all the secondary and intermediate outcomes of interest at

baseline and follow-up?


METHODS


16

At the institutional level, two measures of college environmental change were piloted. First, progress towardsa tobacco-free environment, determined via an audit of FE college policies and structured observations atboth intervention and comparison settings pre and post intervention. Second, staff commitment to smokingprevention and delivery of smoke-free messages, assessed via student survey items at follow-up.

At the community level, the availability of tobacco to students aged < 18 years from local retailers wasassessed via a pre- and post-intervention mystery shopper audit of retailers within 1 km of intervention andcomparison sites and items on the student follow-up survey.

Data sourcesQuantitative data were collected at baseline (September 2014) and follow-up (September 2015) via thefollowing methods: student surveys, college policy audits, structured observations of the college environmentand mystery shopper visits. These methods are described in detail in turn below. All baseline data collectionwas completed prior to randomisation (October 2014). The investigator team was unblinded to allocation atfollow-up, but all fieldworkers remained blinded throughout the study.

Student surveysStudents were eligible to participate at baseline if they were aged between 16 and 18 years on 1 September2014 and had enrolled into FE studies in the 2014–15 academic year (i.e. they were new FE students,aged 16–18 years). Students who were older or younger than 16–18 years and completed the survey wereexcluded from analyses.

As this was a pilot trial, a power calculation was not required. The estimated sample size at baseline of2500 students in six FE settings was chosen to provide some information on variability within and betweensettings at baseline and follow-up. This sample was chosen to indicate the likely response rates and permitestimates [with 95% confidence intervals (CIs)] and intracluster correlations (ICCs) of weekly smokingprevalence in advance of a potential effectiveness trial involving a larger number of clusters (colleges)and students.

At baseline and follow-up, the consent form and survey were completed using an e-questionnaire for easeof delivery and completion in all areas of college, including social spaces without desk access, with papercopies available if necessary (e.g. requested by student, because of technical problems, etc.). Use ofincentives is considered fair recompense for time in work with young people.44 Here, student participationwas incentivised via prize draws for an iPad (Apple Inc., Cupertino, CA, USA) and shopping vouchers atboth baseline and follow-up.

In the first week of September 2014 (i.e. the first week of the new term), students at each participating FEsetting that used an e-mail system (four out of six institutions) were contacted directly via their new e-mailaccount and asked to complete the baseline survey directly via a weblink, they were also sent a remindere-mail 3 weeks later. Those students who did not complete the survey online directly via this e-mail link,or who attended an institution without a student e-mail system, were given multiple opportunities to completethe e-questionnaire on site during September 2014 via: (1) timetabled classroom periods dedicated to surveycompletion, in which students used either college computers, their own devices (laptop, tablet computer orsmartphone) or Google Nexus tablet computers (Google Inc., Mountain View, CA, USA) provided by thefieldworkers; or (2) informal data collection sessions (using Google Nexus tablet computers and/or quicklayered response codes) in common areas at break periods. Hard copies were available as a backup (e.g. if theinternet connection was too slow or could not be accessed temporarily) and were entered online oncefieldworkers returned to the office. All baseline collection occurred between 1 and 30 September 2014.Detailed contact information (name, personal e-mail and mobile phone number) was collected at baseline tohelp track students who left or were on work-based placements at follow-up.



17

The 1-year follow-up took place when students had begun the next college year in September 2015 andused the same methods (i.e. e-mails to all students with a college e-mail account, classroom sessions,informal data collection sessions in common areas). To increase response rates, participants were alsotelephoned up to three times in September 2015 to collect follow-up survey data. All follow-up datacollection occurred between 1 and 30 September 2015. Participants who left college or were unable to becontacted were lost to the trial at follow-up.

After the follow-up survey, all students were contacted and asked to provide a saliva sample in order toexamine the reliability and validity of the self-reported smoking outcome via cotinine and anabasinetesting. However, it was not feasible to recruit sufficient students to provide these samples, despite the useof £10 vouchers as incentives, and we do not report the results of this validation substudy here.

College policy auditsInstitutional policies were obtained in September 2014 (baseline) and September 2015 (follow-up).Researchers requested all relevant tobacco policy documents directly from the management teams at eachparticipating FE setting. Data were extracted into an online pro forma to capture the following informationabout each setting:

l whether the institution has a specific tobacco policy in place or if tobacco use is covered in otherinstitutional policies

l the date of the policy and how often reviewedl whether or not students and/or staff are allowed to smoke or use e-cigarettes on site according to

current policyl whether or not college policies make provisions for cessation services and other ‘quit resources’ on sitel other details, including whether or not careers events, funding and financial connections are covered

by existing institutional policies.

Structured observations of the college environmentOne researcher completed a structured observation of each college environment (n = 6) at baseline andfollow-up using a tablet computer. These observations aimed to assess student and staff practices(e.g. smoking outside of designated areas, where e-cigarettes are used, etc.) and the extent to which thephysical environment at each site communicated tobacco-free messages (e.g. through signage) and/orsupported institutional policies (e.g. directed people to designated smoking areas off site). Photographs werealso taken on the tablet computers to illustrate the information recorded in the observational schedule.

Mystery shopper visitsTobacco availability to students aged < 18 years was assessed at baseline and follow-up using a mysteryshopper audit of all retailers within 1 km (i.e. within a 10-minute walk) of all six participating FE sites.These were the shops that were in the target area for intervention. The aim of the visits was to assesswhether or not local shops were compliant with the restriction of tobacco sales to under-18-year-olds atbaseline and/or follow-up, and to explore differences over time overall and by arm. The protocol for themystery shopper activity was developed with input from Caerphilly Trading Standards officers in line withbest practice guidance for test purchasing methods.

Mystery shopper exercises were carried out by two young people aged 17 years (one in 2014, one in 2015)who were (1) aged < 18 years and (2) not students at any of the sites included in the study. At both baselineand follow-up the mystery shoppers were male. The shoppers were accompanied by an adult fieldworker,who remained outside the shops and out of sight. The shopper entered the shop and asked to purchasecigarettes while the fieldworkers waited outside. They then completed an online, standardised checklist withthe fieldworkers that covered type of shop visited, whether or not they were able to purchase cigarettes andthe presence of age restriction warning posters or materials in the shop. There was one instance at baseline(none at follow-up) in which the shopper was unable to recall whether or not there was additional signage inthe store regarding smoking or age restrictions.

METHODS


18

A total of 18 shops were followed up at both baseline and follow-up; seven shops were also visited onlyonce (either baseline or follow-up) because of closure or because their identity could not be verified atfollow-up (e.g. name changed). In Chapter 3 we therefore report the results when data are available forshops at both baseline and follow-up (n = 18).

Statistical analysesThe primary aim of the pilot trial was to assess feasibility and acceptability, and gather data to plan afuture definitive trial. This included estimating rates of eligibility, recruitment, retention at the 1-yearfollow-up, as well as the acceptability, reliability and rates of completion of pilot primary and secondaryoutcome measures.

The eligibility, recruitment and retention rates for colleges and students are summarised using aConsolidated Standards of Reporting Trials diagram (see Figure 2). The data collected for trial participantswere summarised by trial arm and combined across arms. The aim was to examine the acceptability ofpotential primary and secondary outcome measures for a future trial, as well as describing the baselinecharacteristics of participants. The percentage of missing values was reported for all variables. Categoricalvariables were summarised using the percentage in each category. Numeric variables are summarisedwith the mean, standard deviation (SD) and a five-number summary (minimum, 25th centile, median,75th centile, maximum). We present mean and median values to examine the shape of each distribution.All analyses used intention-to-treat populations.

We used chi-squared and t-tests to examine differences between students who did and did not providedata at baseline and follow-up. Cronbach’s alpha was calculated to examine the internal consistency ofmeasures. Multilevel logistic regression models adjusting for baseline weekly smoking status, age, gender,residence with an employed adult, ethnicity and educational attainment [five or more General Certificateof Secondary Education (GCSE) at A*–C] were used to conduct exploratory effectiveness analyses. Allstatistical analyses were conducted with Stata® version 13 (StataCorp LP, College Station, TX, USA).

Data from the mystery shopper, policy audit and environmental observations were collated and analysedusing Excel spreadsheets. For the mystery shopper survey, shops visited at both baseline and follow-upwere matched in the spreadsheet and the data for each survey item (e.g. whether or not cigarettes weresuccessfully purchased) compared. For each site, items on the policy audit were compared at baseline andfollow-up, with differences between intervention and comparison sites scrutinised. Similarly, environmentalobservation items were compared between intervention and control site, noting any changes betweenbaseline and follow-up.

Economic analysis

We piloted a brief health service use survey for student completion and the EQ-5D-5L (pilot secondaryoutcome) to record preference-based health-related quality of life.39 These measures were piloted becauseit was anticipated that, if feasible to collect data from students in these settings, they could be used inany subsequent Phase III trial to measure any short-term impact of smoking on health-care use and/orhealth-related quality of life.

Trial registration, governance and ethics

The study was funded by the NIHR PHR programme (13/42/02). The trial protocol was registered with CurrentControlled Trials (ISRCTN19563136). The study was overseen by a Trial Steering Committee, comprising anindependent chairperson (Professor Paul Aveyard, University of Oxford) and three other independent members(Professor Angela Harden, University of East London; Professor Rob Anderson, University of Exeter; andDr Julie Bishop, Public Health Wales). The Trial Steering Committee met every 6 months throughout the study



19

(three times in total) to examine the methods proposed and monitor data for quality and completeness.With the agreement of the Trial Steering Committee and NIHR PHR co-ordinating centre, a separate DataMonitoring and Ethics Committee was not established because this was a pilot trial with no interim analysis.

The study was approved by the Cardiff University School of Social Sciences Research Ethics Committeeprior to recruitment and data collection commencing. The validation substudy (saliva testing) was approvedby the School of Medicine Research Ethics Committee, liaising with Cardiff University Human Tissue Actmanagers as necessary. The pilot trial protocol was approved by the South East Wales Trials Unit and theTrial Steering Committee.

METHODS


20

Chapter 3 Results

In this section of the report, we first describe the student response rates at baseline and follow-up and thecharacteristics of the students participating in the study. The findings related to the primary objective are

then presented, which was to assess the prespecified ‘progression criteria’ considered to represent evidenceof feasibility and acceptability of the intervention and trial methodology. We then present student, staff andfacilitators’ experiences of the intervention and trial process. Finally, we report the quantitative data analysesexamining the pilot primary and secondary outcomes, including the pilot economic outcome measure andthe pilot intermediate outcome measures.

Description of pilot trial sample

Six colleges were assessed as being eligible to participate in the trial and participated in the study, includingrandom allocation (three intervention and three control; Figure 2). Colleges could not always know or provideexact ages of enrolled students before data collection, which is discussed further in Chapter 4.

123 ineligible as aged < 16 or > 18 years

Four consented, but provided no data


One consented, but provided no data


Four removed as fieldworker error

255 (19.4%) participated238 (18.1%) analysed

597 (57.1%) participated

470 (45.0%) analysed660 (50.1%) analysed


Six colleges invited to participate

Six colleges entered into random allocation

Baseline data collection (September 2014)

1-year follow-up (September 2015)

Three colleges allocatedto normal practice

Estimated number ofeligible students

(n = 1045)

Three colleges allocatedto Filter FE

Estimated number of eligible students

(n = 1318)

217 (20.7%) participated174 (16.7%) analysed

723 (54.8%) participated

FIGURE 2 The Consolidated Standards of Reporting Trials diagram.



21

Flow of participants in the pilot trialOf the target population of 2363 students at baseline, 1320 (55.8%) participated in baseline datacollection. Of those 1320 participants, five students (0.4%) provided consent information but did notanswer any questions on the survey and 185 (14.0%) were ineligible because they were aged < 16 yearsor > 18 years. Of those excluded as a result of age, around half (49.7%) were aged ≥ 21 years. Theremaining 1130 students, 470 in the control arm and 660 in the intervention arm, equated to a baselineresponse rate of 47.8% based on the estimated total population of new students aged 16–18 years atthese college campuses.

At the 1-year follow-up, 472 (35.7%) baseline respondents participated in the repeat survey. Fifty-six ofthese participants (12.0%) were students who were ineligible to take part in this study at baseline becausethey were aged < 16 years or > 18 years. Four participants’ responses were removed because of concernsabout the quality of data collected on 1 day at one site, based on fieldworker reports (Laurelton College).Out of the 2363 potentially eligible students at baseline, 412 (17.4%) students who were eligible toparticipate provided valid survey data at baseline and at the 1-year follow-up. This comprised 238 studentsin the intervention arm and 174 students in the control arm.

Student characteristicsThe categorical baseline characteristics for eligible and non-eligible participants are summarised andcompared in Table 5.

TABLE 5 Summary of categorical baseline demographic characteristics according to sample eligibility

Variable

Baseline data: distribution over categories by eligibility, %a

Eligible (aged 16–18 years)(n= 1130)

Not eligible (aged < 16 or> 18 years) (n= 185)

Gender

Missing 0 0

Male 37.1 38.9

Ethnicity

Missing 0.1 0

White British 92.8 91.4

White not British 0.9 2.7

Mixed race 3.1 1.1

Asian or Asian British 0.9 1.6

Black or black British 1.2 1.1

Other 1.1 2.2

Is the adult you live with in paid work?

Missing 2.7 34.5

Yes 76.8 46.0

No 14.2 15.1

Not sure 6.4 4.3

Studying full or part time?

Missing 1.3 2.7

Full time 95.3 85.4

Part time 3.5 11.9

RESULTS


22

The eligible participants (i.e. those aged 16–18 years) who completed the baseline survey werepredominantly female (62.9%), white British (92.8%), living with an adult in paid work (76.8%), studyingfull time (95.3%) and had five or more GCSEs (74%). They were enrolled on a wide range of courses,although the two most common FE pathways reported by participants at baseline were AS (AdvancedSubsidiary level)/A-level courses (34.5%) and BTEC courses (32.9%).

The distribution of gender was similar in eligible and non-eligible participants. Ethnic group was evenlydistributed across the two groups (eligible and non-eligible participants), with few non-white participants.Non-eligible participants were more likely than eligible participants to study part time, have a full-time joband be studying for an access level course, but less likely to have five or more GCSEs and study AS/A levels.

TABLE 5 Summary of categorical baseline demographic characteristics according to sample eligibility (continued )

Variable

Baseline data: distribution over categories by eligibility, %a

Eligible (aged 16–18 years)(n= 1130)

Not eligible (aged < 16 or> 18 years) (n= 185)

Working full or part time?

Missing 1.5 3.2

Full time 6.7 12.4

Part time 26.4 27.6

I do not work 65.4 56.8

Do you have five or more GCSEs?

Missing 0.9 2.2

Yes 74.0 51.9

No 19.0 33.0

Not sure 6.1 13.0

Qualification(s) studying at college

Missing 1.2 2.7

AS/A level 34.5 8.11

BTEC 32.9 29.7

Access level course 7.4 12.4

GCSE 5.1 7.0

Other vocational award, certificate or diploma 4.1 4.9

Welsh Baccalaureate 0.4 0

Other 5.7 15.1

Apprenticeship 1.4 3.7

Essential skills 2.4 5.4

HNC 0.2 0.5

HND 0.4 2.3

NVQ 4.4 8.2

AS level, Advanced Subsidiary level; HNC, Higher National Certificate; HND, Higher National Degree; NVQ, NationalVocational Qualification.a Three participants did not provide their age, but provided other data.



23

The categorical baseline characteristics of eligible participants by trial arm are summarised in Table 6.Gender was not evenly distributed across the trial arms: the control group comprised 44.7% males and theintervention group 31.8% males. Participants in the control group were older than intervention groupstudents, with 15.1% of participants at control sites aged 18 years, compared with just 8.2% at interventionsites. Ethnicity was evenly distributed, with very few non-white participants in either arm. Control groupparticipants were more likely to study part time and live with an adult in paid work, but less likely to have fiveor more GCSEs, and be studying for AS/A-levels than intervention group participants.

TABLE 6 Summary of categorical baseline demographic characteristics by trial arm

Variable

Baseline data: distribution of categories by trial arm, %

Control (n= 470) Intervention (n= 660) Overall (N= 1130)

Gender

Missing 0 0 0

Male 44.7 31.8 37.1

Age (years)

Missing 0 0 0

16 60.4 69.1 65.3

17 24.5 22.7 23.4

18 15.1 8.2 11.0

Ethnicity

Missing 0.2 0 0.1

White British 91.3 93.9 92.8

White not British 0.4 1.2 0.9

Mixed race 2.6 3.5 3.1

Asian or Asian British 1.5 0.5 0.9

Black or black British 2.6 0.2 1.2

Other 1.5 0.8 1.1

Is the adult you live with in paid work?

Missing 4.1 1.7 2.7

Yes 68.9 82.4 76.8

No 18.7 10.9 14.2

Not sure 8.3 5.0 6.4

Studying full or part time?

Missing 0.9 1.5 1.3

Full time 93.4 96.7 95.3

Part time 5.7 1.8 3.5

Working full or part time?

Missing 1.3 1.7 1.5

Full time 7.7 6.1 6.7

Part time 23.8 28.2 26.4

I do not work 67.2 64.1 65.4

RESULTS


24

Progression criteria assessment

The first objective was to assess if prespecified feasibility and acceptability criteria were met to determinewhether or not to progress to a larger, Phase III trial to examine effectiveness (RQs 1–6). Interventionfeasibility and acceptability addresses RQs 1–3: did the intervention activities occur as planned in (at least)two out of three intervention settings?; were the intervention activities delivered with high fidelity acrossall settings?; and was the intervention acceptable to the majority of FE managers, staff, students andthe intervention delivery team? The questions regarding the acceptability and feasibility of trial methods(RQs 4–6) are then addressed in Trial feasibility and acceptability.

Intervention feasibility and acceptabilityFirst, each of the five intervention components is considered in turn to understand whether or not theywere delivered as planned (RQ 1) and with high fidelity across multiple settings (RQ 2). At the end of thissection, the question of overall acceptability is addressed briefly (RQ 3) prior to presenting participants’views in more detail in Process evaluation: participants’ experiences.

Prevention of the sale of tobacco in local shopsWith the aim of restricting the local availability of tobacco at each intervention setting, the Filter FE interventionincluded a community-level component targeting local retailers. The aim was for the intervention manager to

TABLE 6 Summary of categorical baseline demographic characteristics by trial arm (continued )

Variable

Baseline data: distribution of categories by trial arm, %


Do you have five or more GCSEs?

Missing 1.1 0.8 0.9

Yes 63.0 81.8 74.0

No 26.4 13.8 19.0

Not sure 9.6 3.6 6.1

Qualification(s) studying at college

Missing 1.7 0.8 1.2

AS/A-level 24.8 41.4 34.5

BTEC 30.4 34.7 32.9

Access level course 7.2 7.6 7.4

GCSE 8.3 2.9 5.1

Other vocational award, certificate or diploma 5.3 3.2 4.1

Welsh Baccalaureate 0.6 0.2 0.4

Other 8.5 3.6 5.7

Apprenticeship 1.5 1.4 1.4

Essential skills 3.8 1.4 2.4

HNC 0 0.3 0.2

HND 0.2 0.5 0.4

NVQ 7.4 2.3 4.4

HNC, Higher National Certificate; HND, Higher National Degree; NVQ, National Vocational Qualification.



25

contact all shops selling tobacco within 1 km of the intervention setting (i.e. within a 10-minute walk).This component was intended to be delivered immediately post randomisation (early October 2014) via letterssent to these retailers to inform them that a new smoking prevention project (The Filter FE) was taking place attheir local FE institution. Posters, stickers and other materials about The Filter and ASH Wales were also tobe distributed to these shops to provide additional information to them and their customers about the legalage for purchasing tobacco products and the requirements to produce statutory identification to purchasetobacco. Owing to the location of the intervention settings in rural and suburban areas, in total therewere only five shops within 1 km of all the intervention sites; one rural site had no tobacco retailers within1 km (Laurelton College).

The results indicate that the criteria for RQs 1 and 2 were not met for this component at any of the threeintervention sites. Despite the low number of retailers, the intervention team checklists were not recordedconsistently and intervention team members reported that resources were not sent out to these retailersas planned. Various factors appeared to contribute to this, but the most common factors appeared to bethe relatively short time available to identify and work with local community settings, as well as poormanagement of intervention implementation. There was also no evidence of changes in practices from themystery shopper audit at follow-up in September 2015. In two out of the three shops close to interventionsites that were audited by a mystery shopper at baseline and follow-up, it was possible to buy cigarettesat follow-up but not at baseline (i.e. local availability of tobacco to students aged < 18 years may haveincreased; the findings of mystery shopper audit are reported in Pilot intermediate outcome measures).There was some observational evidence that age restriction signage increased in local shops betweenbaseline and follow-up, although this was consistent across shops in both intervention and control arms.

Policy review to promote a tobacco-free environmentFive policies were audited and recorded at baseline. In one (control) site, Afonwood School, there was nowritten policy, but if students were caught smoking on site, a letter was sent to their guardian (a copy ofthe letter was obtained). With the aim of restricting opportunities for smoking and promoting non-smokingas the norm at each intervention setting, it was intended that the intervention manager would developnew whole-campus tobacco-free policies with each intervention setting. This component was intended tobe delivered immediately post randomisation (early October 2014) by the intervention manager whowould work with FE managers at each intervention setting to review their institutional policies using thetobacco-free campus guidance developed by ASH Australia.34

Overall, there was little evidence that the intervention delivery team worked with intervention sites tosupport the introduction or modification of smoking policies. As with the prevention work targeting localshops (above), the intervention team checklists were not recorded consistently and intervention teammembers reported that this intervention component was not implemented as planned, again becauseof the short run-in time and poor management of the intervention implementation. The ASH Australiathree-stage tobacco-free campus guidance was not used as intended at any site.

Analysis of institutional policy documents at follow-up found limited evidence of changes made to policies:at one intervention site the smoking policy was amended to include e-cigarettes, but both smoking ande-cigarettes were still permitted in designated ‘smoking areas’; fieldworker notes at one interventionsetting recorded that one college had adopted a ‘campus-wide no smoking policy’ between baseline andfollow-up but there was no written policy and it was not possible to attribute any changes directly to thisintervention component. At the end of the intervention, none of the available school/college policiesexplicitly addressed issues (such as tobacco industry sponsorship, funding and gifts or tobacco advertisingand sales on site), which are suggested in the ASH Australia guidance.

Further education staff trainingIt was intended that a minimum of one, two or four staff training sessions would be delivered at theintervention settings, proportionate to institutional size. Staff training was delivered at all intervention sitesby experienced training officers employed by ASH Wales on The Filter youth project. However, in practice,

RESULTS


26

the number of sessions delivered was driven by demand in each site and only three sessions were deliveredin total, one in each site (i.e. lower than intended at the two FE college campuses). All sessions wereattended by at least six staff, with a total of 28 staff attending overall, who were observed to be engagedand interested in the topics. At Valeside College, the large FE campus in the intervention arm, attendancewas lowest (six people); a second training session had been arranged but was cancelled because of a lackof staff interest. ASH Wales staff used its own standard evaluation forms at the end of each staff training.However, these did not collect additional information we required (e.g. staff role, previous training, contactwith students, etc.) and, therefore, were not collated or used in the analysis.

The content of the training was agreed between intervention sites and the intervention delivery team asplanned, drawing on the different teaching resources developed and piloted by ASH Wales in schools andother youth settings for The Filter youth project. All of the FE settings chose the staff training resourcesdeveloped by ASH Wales on e-cigarettes, and these sessions covered the same topics using the samecombination of Microsoft PowerPoint® 2013 (Microsoft Corporation, Redmond, WA, USA) presentation,group exercises and discussion with high fidelity across all settings.

Social mediaObservations of the college websites provided evidence that links to The Filter social marketing campaigns,and its advice and support services (e.g. The Filter text/instant messaging services), had been publicisedon the institutional websites as planned. The intervention team also reported some use of college socialmedia channels to provide more information about The Filter services and its smoking prevention messages(e.g. via retweeting on Twitter or sharing on Facebook). However, the intervention team member responsiblefor social media also reported it was challenging to engage the colleges: they did not have clear systemsand structures for using social media, and did not appear to want to publicise smoking prevention messages.None of the three intervention sites embedded information on their home/index webpage or other frequentlyaccessed webpages/microsites (e.g. online learning portals, e-mail login pages), as intended. Site staff alsoneeded repeated prompting to share information about the intervention on social media.

Youth work activitiesIt was intended that 5, 10 or 15 groups (depending on institutional size) of 10–20 students would takepart in 2-hour sessions provided on site during school/college time. Youth work sessions were delivered ineach of the intervention sites, with a total of 15 youth work sessions delivered overall: 10 at the largestcollege, two at the other college and three in the school sixth form. A total of 190 students were reachedacross the three sites. The number of students who participated in each session varied significantly, from4 to 40. The number of participants per session was highest in the school sixth form sessions, in which80 students attended in total across the three sessions. Although planned to begin in the autumn term(October–December 2014), all the youth work activities took place in the spring and summer terms 2015.

Sessions were organised and run by facilitators from the intervention delivery team as planned, using TheFilter youth work resources to educate and persuade students about the harms of smoking and modelsocial/situational resistance skills. Facilitators agreed the content and length of sessions with staff at eachintervention site and there was no evidence that young people were consulted at any school/college orinvolved in the design of activities.

In two of the three observed sessions, students worked in small groups to create smoke-free healthpromotion campaigns, then presented their ideas and voted on them at the end; the other session wasa more general workshop that the facilitator said was intended to enable participants to explore howperceptions of smoking have changed over time. Participants did not appear to be engaged in the twosessions in which they developed a smoke-free message campaign, some students in the other observedsession were reported to be engaged by some of the shorter, more interactive activities.



27

Was the intervention acceptable to the majority of further education managers, staff,students and the intervention delivery team?In summary, low levels of acceptability were reported by the majority of participants at each pilot setting,meaning that the criteria for RQ 3 were not met. One major barrier to acceptability was the perceptionamong FE staff and students that FE is ‘too late’ for smoking prevention interventions: current smokersreported disengagement and may benefit more from access to cessation services; and non-smokers do notwant to be ‘bombarded’ with more educational messages about smoking in post-16 settings.

Another key barrier appeared to be the institutional cultures of FE settings, which promotes autonomy andstudent responsibility that, in turn, limits the willingness of staff to intervene on issues such as smokingand students’ engagement with such activities. Smoking prevention is not a priority for FE managers andstaff, which further limits the acceptability of a multicomponent smoking prevention intervention such asthis, although they did support external health agencies’ involvement in FE settings. The intervention teamalso found that the proposed smoking prevention activities were significantly less acceptable in collegesettings than in schools and other youth settings for these reasons (see Attitudes towards smoking in thefurther education context).

Participants’ views are presented in more detail in Process evaluation: participants’ experiences to explorethese sources of unacceptability in more depth, in the wider context of student attitudes towards smokingand institutional culture, and alongside other barriers to implementation.

Trial feasibility and acceptabilityThis section addresses the progression criteria relating to trial methods and RQs 4–6: was randomisationacceptable to FE managers?; did (at least) two out of three colleges from each of the intervention andcontrol arms continue to participate in the study at the 1-year follow-up?; and do student survey responserates suggest that we could recruit and retain at least 70% of new students in both arms in a subsequenteffectiveness trial?

Interviews identified that randomisation was acceptable to FE managers. Recruitment and baseline surveystook place as planned, with all six FE settings retained at the 1-year follow-up. No college managerobjected to the use of randomisation to intervention and comparison groups.

Recruitment and retention of students was challenging, despite the use of the multiple methods andincentives. Student survey response rates in this study did not suggest that it would be feasible to recruit andretain at least 70% of new students to a cluster RCT in FE settings without further methodological work.

The target population of 2363 enrolled students comprises some ineligible students as a result of settingshaving incomplete enrolment data on students aged > 18 years, and is therefore the incorrect denominator.In September 2014, 1320 students out of an estimated total sample of 2363 participated in the baselinesurvey. Of these 14.0% (n = 185) were ineligible as they were aged < 16 years or > 18 years and fivestudents provided no data, leaving a baseline sample of 1130 (47.8%). Although this equates to a responserate of < 50%, the number of potentially eligible students at baseline (n = 2363) was provided by eachinstitution and overestimates the actual number of new students aged 16–18 years in that setting and,therefore, underestimates the true response rate. If data on the number of eligible students were available,the correct denominator could be used and the baseline response rate would be higher.

In September 2015, 412 eligible students completed the follow-up survey (36.5% of baseline respondents;17.4% of all potentially eligible students at baseline), which is significantly below the target responserate of 70%.

RESULTS


28

Process evaluation: participants’ experiences

Our second objective was to explore participants’ experiences of the intervention and pilot trial methods(RQs 7–10). We conducted semistructured observations of staff training sessions and youth work sessionsdelivered by The Filter team, focus groups with students and staff at each intervention site, and interviewswith FE managers at intervention and control sites and The Filter staff who managed and delivered theintervention to obtain their views on the barriers to, and facilitators of, implementation. This section reportsthe results of analyses of these data to address RQs 7–10. It is divided into four subsections: Attitudestowards smoking in the further education context reports common student and staff views about smokingand transitions to FE, and how these interrelate, to understand the societal and institutional context inwhich The Filter FE intervention was being piloted; Barriers to acceptability and implementation builds onthese analyses to report the key barriers to implementation, including the sources of unacceptability, ofsmoking prevention activities in this context; Implementation of intervention components and contextualvariation describes participants’ experiences of each of the five intervention components separately; andResearch methods: feasibility and acceptability reports the acceptability and feasibility of the trial design andresearch methods.

Attitudes towards smoking in the further education contextDuring focus groups and interviews, FE students and staff reported their attitudes towards smoking, andthe extent to which they perceived the uptake of smoking to be a problem for this age group and/orwhether or not prevention is a priority for FE settings to address. These focus groups therefore providedrich contextual data to understand the environment and systems within which the planned interventionactivities were piloted. Through understanding these FE settings, and student and staff norms aboutsmoking, we are able to more fully theorise the sources of unacceptability and other barriers toimplementation described in Barriers to acceptability and implementation.

The importance of the principles of freedom, personal responsibility and self-determination in FE settingswas a recurring theme emerging across focus groups with both students and staff at all sites. Participantsoften explained this by contrasting FE settings with more constraining school environments, where staff areconsidered to be ‘stricter’.

Three cross-cutting themes emerged that exemplified this desire for freedom, personal responsibility andself-determination, as well as the power of wider societal norms about smoking, to limit the acceptabilityand feasibility of smoking prevention activities in the context of FE settings. These cross-cutting themes,described below, are:

1. the hegemony of anti-smoking norms2. the FE transition and liminal identities3. smoking as an individual choice in FE.

The hegemony of anti-smoking normsAnti-smoking attitudes were common among FE students and staff, including college managers. Participantswere aware of the health harms of smoking, felt these were universally understood and agreed with thesmoking ban in public places. Reflecting on national trends in smoking rates and attitudes about smoking inpublic, students realised that smoking was now much less socially acceptable. For example, one studentexplained:

Back in the olden days, it’s just everyone used to smoke didn’t they, and they used to think itwas normal.

Athervale School, student focus group



29

The minority of students who smoked were typically guarded about their smoking status and felt thatpeople made negative judgements about them specifically because they smoked. For example, onestudent stated:

I’ve met loads of people that smoke that aren’t as bad as you think they are. [. . .] People arestereotypical about smokers. Because they think they’re, they’re doing it because they want to lookcool and stuff, but it’s nothing to do with that really.

Laurelton College, student focus group

Both non-smokers and smokers frequently used words like ‘dirty’, ‘smelly’, ‘stink’ and ‘disgusting’ to referto smoking. One group of non-smoking students at Athervale School said the taste and smell was‘disgusting’ and ‘gross’; another non-smoking student from Valeside College said, ‘I just find it pointless’.A staff member felt that FE students now frequently commented explicitly to each other about the smell,saying that:

If someone comes in and they’ve been smoking, they’ll pass a comment as well, you know, if it’s otherstudents, they’ll say ‘Oh go away, you stink!’, you know [. . .] Because there’s less people smoking soit’s more noticeable.

Valeside College, staff focus group

Smokers also reported this process and how they felt judged by some non-smokers at their college:

I don’t like the, not the snobbiness but the, kind of ‘oh it’s a disgusting, filthy thing’ [. . .] because it’san addiction isn’t it? But we don’t force anybody to smoke, we don’t say ‘yeah, oh you should smoke,why not’.

Valeside College, student focus group

Students who smoked occasionally – ‘socially’ – also appeared not to want to identify themselves as‘smokers’ for this reason:

I don’t know if I’d, like when I fill in forms I don’t know if I consider myself a smoker, because I dowhen I drink, but I don’t know, about half one [. . .] My friends always joke and say that I am. Which Iguess I am because I do smoke, I suppose, if you do smoke sometimes I guess you are down as beinga smoker, so.


Staff were also aware of the negative connotations of being a ‘smoker’ and described attempting toexplain to students on some courses (e.g. health and social care) that this would likely be the case forthem in their future workplaces.

The further education transition and liminal identitiesReflecting the power of these social norms about smoking, students also perceived smoking as a behaviourthat brings particular responsibilities and requires them to show ‘respect’ for others, particularly children.Through sensitivity to social norms, and in the context of recent bans on smoking in public, the transitioninto FE enabled some students to demonstrate how they are responsible adults and, thus, differentiatethemselves from (school) children. For example, one student explained:

I think everyone just respects the fact that some people don’t smoke and don’t want smoke in theirface like, so it’s just that respect. It’s like if I’m walking down the street and there’s a woman and herkid in a pram or something walking by and I’m smoking, I’ll cross the street because I don’t want toblow that in the kid’s face like. It’s just out of respect ‘cos if they don’t smoke then, they shouldn’thave to have second-hand smoke.


RESULTS


30

Although many of the students who identified as smokers (not social smokers) stated that they begansmoking at a relatively young age, they still emphasised consideration for the welfare of non-smokers,particularly children, and supported bans on smoking in public. This suggests that during transition to FE,attitudes and actions relating to smoking helped to signify students’ co-occurring transition to adulthoodand self-determination. This was also demonstrated in several participants’ indications that they would notbuy tobacco for younger students, but may sell on to those in year 10 or 11 (aged 14–16 years) – ‘if Iknow them but not people in like year seven and eight [aged 11–13 years] and that’ – reflecting their ownexperience of commencing smoking in year 10 or 11.

The liminal status resulting from transition was further apparent in the characterisation of children andyounger students as others. For example, when discussing e-cigarettes, students cited concern over thepotential for e-cigarettes to act as a gateway to using tobacco, particularly for children. As one student put it:

Someone will use e-cigarettes as a start off before actually smoking properly, especially youngerchildren tend to have them and then move on to the actual cigarettes.


As well as acknowledging their responsibilities to protect younger people, students saw themselves aspotential role models to children, typically differentiating themselves from children who have littleautonomy and require protection from the effects of smoking.

Staff discussions of student smoking behaviour reflected this perception of students in transition andillustrated a sense of responsibility to help students negotiate this while respecting their personaldecision-making and nurturing their autonomy. As one staff member explained:

It’s their choice to smoke. Now, either we can give them the messages about not smoking, but youknow, we’re not, we don’t preach to them about not smoking, our main message to them is if you’regoing to smoke, from my point of view, it’s making sure that they do it in the correct way, smokingthe correct things, not illegal substances, and they’re in the right places. As long as they’re doing that,we, we’re off their back really.

Laurelton College, staff focus group

Both students and staff also discussed smoking as something young people can do to rebel, to asserttheir autonomy, especially in the face of opposition from authority figures. One student who smokedarticulated this as:

It’s the trying to push people to stop that tends to make them think, no, balls to you.Valeside College, student focus group

This resistance was expected to further limit how receptive students may be to advice and education in thiscontext. As one staff member described:

I think it’s just, they’ve hit that particular age and, and as I said, the adult in them is trying to comeout and they do not, they’re not receptive.


Recognition of the liminal nature of their current status meant that students were also anticipatingimpending transitions, for example starting higher education or entering work, which affected their viewsabout their future patterns of smoking. When asked, most non-smokers did not think that they wouldstart to smoke in the future and most current smokers did not believe they would continue smoking intotheir twenties, suggesting associations between smoking and perceived life stage.



31

Smoking as an individual choice in further educationA recurrent theme was that smoking and quitting are choices that individuals make. As one studentexplained, smoking is ‘their choice’, with staff also reflecting this dominant view of autonomy and personalresponsibility as embedded in FE. A staff member at Laurelton College explained:

As long as people are informed of what they’re doing. You know, they’ve got the information.That’s their choice.


This meant that, in recognition of wider societal anti-smoking norms, students typically talked in termsof deciding to smoke as a ‘bad’ choice or not to smoke as a ‘good’ choice. For example, one studentdescribed with admiration how a friend chose not to smoke even when other friends did:

He turned round and said he didn’t like it at all. He just didn’t do it. And walked away from them.It’s a good choice isn’t it?


Consistent with their belief that smoking is a choice, students and staff framed discussions about perceptionsand behaviours relating to smoking in the context of rationally weighing up information on risk and decidingif this information was perceived as reliable and trustworthy. Students (and staff) frequently described whatinfluences their smoking behaviour, including financial considerations, peer and family behaviour.

Reflecting the enhanced autonomy of the life stage, staff also believed that increased financialindependence was significant in student decisions to spend money on smoking if they wanted to:

There’s an economic factor as well, because if you’re a student who’s suddenly getting your own EMA[Education Maintenance Allowance] and getting 30 quid a week, all of a sudden you can afford to goand buy that packet of cigarettes. It might last you 2 weeks, it might last you a day, but you’ve actuallygot the money and the buying power, whereas you might not have had that when you were 15.


This further contributed to the sense that smoking was an individual choice in FE, in which decisions tosmoke are made through a rational, consumer framework of weighing information about costs, personalpreferences and risks to their health and image.

Some students and staff did recognise wider influences beyond the individual’s knowledge and attitudes,particularly peer pressure and the home environment. However, these ‘social things’ were seen asinfluences occurring earlier in the life course:

[Smoking] it’s a social thing and I don’t know many people who smoke so, I think that’s, I, my parentssmoke, my mum and my dad, but I think it’s just, I don’t know, my mates don’t smoke so, I’ve never reallygot that urge to social smoke with my mates so. I think it’s just who you’ve associated yourself with really.


Of all the students who smoked, only two perceived that they had been bullied into smoking by theirpeers, contrasting with the presentation of smoking as a personal decision. However, both studentsexplained that this had happened when they were ‘younger’, in secondary school, when other students‘forced’ them to smoke. One explained:

Truthfully I got bullied into it when I was younger. When I was like 11. I think it was 11 like. Smokingwas, you just think you’re cool as well don’t you, in front of your mates and stuff, but no, not really.I don’t like doing it.


RESULTS


32

Students refuted the suggestion that they were still subject to such peer pressure in FE and typicallyperceived that they could now resist peer influences. In keeping with the perception of smoking as apersonal choice, students and staff were also firmly convinced that cessation is primarily a matter ofself-control. As one student who smoked explained:

If you want to give up you need will, you need willpower. And I just haven’t got any. At all.Valeside College, student focus group

This viewpoint was also shared by non-smokers:

Chances are though a lot of people who smoke they have no intention of quitting so you can’t reallyhelp them, if they don’t want to help themselves.

Athervale, student focus group

Finally, although ‘stress’ was recognised as contributing to smoking and as limiting chances of cessation,students and staff did not consider FE environments to be particularly stressful.

SummaryThis section has described the key themes arising from the discussions with FE students and staff regardingsmoking in the context of transition into FE settings. These analyses illuminate the importance ofself-determination for young people at this stage in their life, particularly in how this manifests and isrehearsed by FE students, and is accommodated in institutional processes during this transitional period.With these themes in mind, the next section describes the barriers to acceptability, and implementation of,The Filter FE intervention.

Barriers to acceptability and implementationAs described above (see Intervention feasibility and acceptability), the intervention was not implemented asplanned. This section describes the key barriers to implementation in two categories: first, the interventionmessage and aims; and, second, wider challenges associated implementing interventions in FE settings.

Intervention message and aimsKey challenges in implementation were rooted in low levels of acceptability of a smoking preventionintervention in FE settings that, as described in Attitudes towards smoking in the further educationcontext, emphasise personal responsibility, autonomy and individual choice. This institutional culture limitsthe potential and willingness to implement The Filter FE intervention as planned, as well as limitingstudents’ engagement with the activities that occurred.

Smoking was not a priority issue for staff or the students who they worked with. Overall, low levels ofacceptability were reported by the majority of students and staff at all sites, including a perception that FEis ‘too late’ for smoking prevention intervention. It was suggested that current smokers would disengageand may benefit more from access to cessation services, whereas non-smokers do not want to be‘bombarded’ with more educational messages about smoking.

These factors led the intervention team to identify that proposed smoking prevention activities weresignificantly less acceptable to students in FE settings than schools and other youth settings. Intervention staffalso perceived that smoking was not seen as an issue, by staff or students, in FE colleges when comparedwith schools, potentially relating to differential exposure. One member of The Filter team suggested that:

I think schools do buy into it more because you are there from nine to three, aren’t you . . . so ifyou’re smoking, it’s very present on a school campus. If you’re in and out for FE, you can come and goas you choose, can’t you? Sixth form, you’ve got that freedom . . . so actually, even if you’re a smoker,the FE tutors might not see smoking on campus to be a massive issue because most people might begoing down the local café . . . park, whatever, to have a fag . . . where it’s much more present . . .



33

even if it’s round the shed or whatever . . . do you know what I mean? . . . It’s easier to hide in anFE setting.

Intervention team member

The data also illustrated staff receptivity to input from an outside agency, particularly on e-cigarettes,about which they felt they had little knowledge. All students (both smokers and non-smokers) reportedhaving sufficient information about smoking tobacco and its effects on health, leading to some reluctanceto engage in further repeated and unnecessary discussion. In contrast, many staff and students wereuncertain how to evaluate the health risks of e-cigarettes, with discussions about e-cigarettes furtherillustrating the frames they felt that they and others used to make decisions about smoking. For example:

There’ve been no long-term studies [of e-cigarettes]. Nobody knows, you know, and you’re back intoa situation of where we were 50 years ago when people thought smoking was fine, you know, andreally healthy for you. But, are we going to say in 10 years down the road, oh my gosh, look whate-cigarettes have done to our kids growing up, they shouldn’t have had them for whatever reason.

Athervale School, staff focus group

Nevertheless, there were positive aspects of the intervention, including FE managers and staff welcomingthe flexibility (e.g. they could choose the topic of the staff training). As one FE manager explained:

I’m always looking to work, you know with external people and whatever they can bring in to ustraining wise, we’ll always welcome that.

Laurelton College, FE manager

It was evident from interviews and focus groups with staff at FE colleges that they, like school staff, have astrong sense of a duty of care, albeit this could manifest in different ways (see Policy review to promote atobacco-free environment on site policy). As one FE manager explained:

Post-16 [FE students] get glossed over sometimes on support work, you know, from outside agenciesum, and, you know, they’re people and students who you can have a big impact on.

Athervale School, FE manager

Implementing interventions in further education settingsAlthough the opportunity to work with outside agencies to deliver an intervention was welcomed (seeIntervention message and aims), there were several cross-cutting challenges to implementing a complexmulticomponent intervention that are specific to a FE context. These centre on management andcommunication systems, intervention timing, timetabling and the nature of courses in FE.

In terms of management and communication, it was difficult for the implementation team to identify whoto contact in each institution, with one FE manager commenting that ‘every organisation is structureddifferently’ (Middledale College, FE manager). This meant that, unlike schools where it might be possibleto directly approach the personal, social and health education co-ordinator, the team found it difficult toidentify who to contact as ‘they’ve all got different tiers’ (ID team member 3). However, one interventionteam member went on to say that it was easier to identify who is in charge of social media in FE collegesthan in schools, because of the presence of dedicated information technology staff:

I think colleges have specific people, whereas [school] sixth forms don’t have like a digital mediaofficer, they have a member of staff who likes tweeting so they kind of tend to put them on it.Whereas in the colleges, they’ll have like one or two people who do the website and the social media,so they’re employed people to do it.

ID team member 3

RESULTS


34

The intervention delivery team members who required direct contact with students and staff (for youthwork and staff training) also struggled to organise convenient times for delivery of sessions as sitesrequired a long lead time to plan such sessions into their timetable. This was compounded by weaknessesin intervention management. For example, a lack of planning and organisation was observed. Theintervention manager also had limited experience of working in FE settings and experienced high staffturnover in his own team during this pilot study.

In terms of the youth work sessions, one FE manager explained that, although it might seem an appropriateactivity for tutorial time, the ‘pressure on our tutors and the timetabling is, is really quite immense now’

(Valeside College, FE manager), meaning that existing pastoral sessions with tutors are already struggling toaccommodate all requirements. They went on to explain:

There’s now only going to be group tutorials, there’s going to be only five of those a year, and one ofthose sessions will be health and well-being, that would include smoking, but it also includes thingslike mental health and other things like that.

Valeside College, FE manager

Moreover, the nature of courses and timing of examinations mean that the academic year is quitetruncated, so the opportunities for intervention are fewer than might be expected. As the Athervale SchoolFE manager explained, the window of opportunity to intervene lasts only from ‘September to the Eastertime, so it’s actually quite a short period of time to do work with students. It’s only 16 weeks’. They wenton to describe their frustration at the organisation of sessions, explaining that with the pressures onstudents’ timetables, activities needed to be planned at least a term in advance.

As a result of the difficulty FE managers experienced in releasing staff and students, intervention teammembers were sometimes presented with smaller groups than they had anticipated, prompting changes tothe delivery style of the session. Lack of prior information on group composition also meant they were alsounable to anticipate the appropriate level to pitch the youth work sessions, as one intervention deliveryteam member described:

I don’t think the students are a problem, like. But it’s also then tailoring to the right level, because youhave, some college students are in college ‘cos they can’t do A-levels, but it’s us knowing beforehandwhat kind of level we’re targeting the work at as well.

ID team member 3

Implementation was also problematic in colleges because of the diverse courses offered, with somestudents absent from campus for extended periods of the academic year on work placement, or attendingcollege only on day release from their workplace. As one FE manager described:

Most of our courses will have some kind of placement element to them, so the target that theresearch was on, we might have, if you got 150 students in that campus, there are probably only80 students on the campus on that given day.


Time pressures and competing priorities made it similarly difficult for intervention staff and FE managers toarrange the staff training. Indeed, time restrictions appeared to be a factor in choosing the sessions; staffwere interested in e-cigarettes but it was also one of the shorter training sessions on offer. One of theintervention delivery team members explained the challenges in setting up the staff training:

It’s been quite hard getting the colleges to organise the sessions [. . .] we did quite detailed attachmentsin an e-mail about the different programmes they [FE managers] can pick from, we made it very clearthat we can be very flexible on times and how many people we have. Um, and we found, I probablyfound that part hard, just because the contact was quite sporadic and actually it was, there were issues



35

with the, probably all three colleges, there were issues with getting them to actually, get, take it, takethe training up, which we were surprised at because we thought, well, they’d agreed to it and, it’ssomething for free, for them.

ID team member 1

The FE managers also reported time pressures as having an impact on participation. As one FE managersaid: ‘it was physically impossible you know, to get all of those staff in there from 9.00 until 5.00’ (LaureltonCollege, FE manager). Another explained that they were keen to be involved but that they were incrediblytime poor:

I think generally students and staff want to be involved, and want to have a go at doing things, it’sjust a question of their, their time is split in so many different ways, and their priority and focus has tobe on teaching, so if they’re missing another lesson from another subject, they become more hesitant,reticent, to do that, yeah?


Further education managers suggested potential options for accommodating staff training, for exampleusing inset days, but, as these are planned up to 1 year in advance, there were some doubts that smokingwould be seen as a priority for such events.

SummaryA key challenge to student and staff engagement was a perception that the messages and aim of theintervention were occurring ‘too late’ to address smoking behaviour. This, combined with institutionalcultures of FE settings that promote autonomy and student responsibility, meant that the willingnessof staff to intervene on issues such as smoking, as well as students’ engagement with such activities,was further limited. Although staff welcomed work with external agencies and perceived a need forinterventions targeting FE students, there are significant challenges to implementing interventions in FEsettings relating to the heterogeneity of these institutions, the diverse courses and timetables beingdelivered, and the short windows of opportunity available for intervention.

Implementation of intervention components and contextual variationAs well as considering the acceptability and feasibility of each intervention component, analysis of qualitativedata led us to question the relevance of the components of the intervention. This primarily related to if theapproach (smoking prevention) was perceived as pertinent in this setting and among this population, andwhether or not all the five intervention components were needed. This section discusses the perceivedrelevance of each intervention component in turn, then describes participants’ views about the impact, reachand awareness of the intervention overall. No significant unexpected consequences of the intervention werereported or observed.

Prevention of the sale of tobacco to further education students aged < 18 yearsShops near to the intervention sites were generally understood by students, staff and FE managers tobe strict about sales practices at baseline (i.e. limited potential for this component to trigger changes).However, students had no trouble obtaining tobacco, which suggests that the focus of this component onpreventing the supply of tobacco from shops near FE settings was misdirected. Some students did reportpurchasing from shops on the way to or from school/college but most commonly they would obtaintobacco from friends old enough to legitimately purchase it and/or from family members (particularlyparents). One student described their father buying cigarettes for them because:

My dad says he’d rather me smoke in front of him than behind his back so that he knows whatI’m smoking.


RESULTS


36

Students also reported sharing and selling tobacco products among themselves, with one stating that:

If one of us haven’t got fags, like our, like our friends are like that, come on, you can have one, sort ofthing like. So no-one goes without.


This notion of reciprocity relates to the main reasons cited by students and staff for students smoking: it isa source of identity and social bonding. Most students who smoked said that they had been asked toeither buy tobacco for younger students or to sell some to them. However, as noted above (see Attitudestowards smoking in the further education context), students developed their own ethical framework forjudging whether or not to do this. When students did go into shops to purchase cigarettes, they indicatedthat they were not deterred now that products are not visible to choose, with one student saying:

We know all the prices off by heart.Athervale School, student focus group

Staff were also aware of alternative access routes to tobacco, with one commenting: ‘there’s alwaysloopholes – they can always get it if they want it’ (Valeside College, staff focus group). Staff mostfrequently perceived that student access to tobacco was through their social networks rather than directlyfrom shops. As one staff member said:

Very, very few students actually go and buy cigarettes or tobacco. It’s, it’s, somebody’s been onholiday or somebody’s selling stuff you know, that grey market . . .


Indeed, staff and students were aware of various illicit supply modes employed by students, with somesuggestion that students were uninterested in the ethics of buying potentially counterfeit products as aresult of more sensitivity to price. For example:

They don’t care. Because again, it’s this thing, of the relevance to the young people. If they go to ashop and spend, I don’t know, £10, £15 whatever a pouch of tobacco is, but they can go and buy itoff him down the road for £5, £6, well which one are they going to do? They’re going for the £5,£6 one.


Staff in this focus group were not inclined to report students for using fake products, but might mentionto students the potential dangers of using non-regulated products. Similarly, in the school sixth form(but neither of the intervention colleges) staff and students were aware of students selling single cigarettesto others for 50p. Staff described this awareness as:

You overhear things ‘do you know who’s selling roll ups’ or ‘so and so’s selling’.Athervale School, staff focus group

However, this practice was unchallenged because ‘you’d have all the abuse and everything else and thenwhen it went further, nothing would get done’. This was supported by a colleague, who added that itcould also ‘impact into your lessons as well if you teach them’. They rationalised this approach by sayingthey would only act if they ‘actually physically see it’ and:

You’re not allowed to search a child’s bag or anything, so it creates a huge commotion doesn’t it, ifyou, it has to involve everybody, it goes like right up the chain. [. . .] but when you think, you know,it’s trivial really isn’t it at the end of the day? A cigarette. You know, for all the commotion. I know itisn’t, but it is, in a day, if you know what I mean.




37

Students were aware that staff were unlikely to try to intervene when sixth form students were selling toyounger students, with one stating that it ‘would be quite hard to stop’ this trade. This was furtherdiscussed:

Student 1: They have no proof that you did do it anyway, so they can’t do anything about it.

Student 2: Unless they’re with the cameras, but they can’t.

Interviewer: So it’s quite hard to stop that kind of like, people passing tobacco.

Student 2: No, because if there was more cameras, people would just leave anyway, and they’ll justlike go somewhere else where there’s a blind spot, they will.


Policy review to promote a tobacco-free environmentAll intervention sites had a policy on smoking before the intervention commenced, reflecting wider societalanti-smoking norms and systemic requirements. Of all the five Filter FE intervention components, the issueof institutional policy on smoking generated most discussion among staff and students in focus groups.Below, we first discuss FE managers’ perceptions of the relevance of this intervention component and itspoor implementation as a result of weaknesses in intervention management and delivery. Second, we goon to discuss staff’s and students’ responses to institutional policies on smoking, including perceptions ofacceptability, as well as strengths and weaknesses of current policies.

There was low awareness of the planned policy review activities but acknowledgement that otherintervention activities had influenced institutional policy processes. For example, one FE manager indicatedthat staff training on e-cigarettes ‘led that into the cognisance, oh we need to think about this’, in termsof how to incorporate them into site policy. The same FE manager also noted that:

One of the researchers said about no smoking signs not being in schools, and actually it wasn’tsomething that we’d thought about. I can’t say that we’ve had any direct outcome, but it’s somethingthat we’ve [the senior management team] got up for discussion again, and to have a look at.


This same manager said it ‘definitely’ would have been useful to have feedback about their policy and‘the right way to do it’. Another intervention site manager said that they had e-mailed their policy to theintervention delivery team and described the background to their desire to carry out a review, saying:

There was consultation done by senior management in this college about 5 years ago and theywanted to ban smoking on campus altogether, but there was huge backlash to that, so then theyobviously introduced designated smoking areas, which has never kind of settled, there is alwayspeople for smoking shelters and there are people against, but that’s why I thought to get an externalbody in, it would be interesting to get view and some extra training on that as well. [. . .] We haven’tchanged the policy, um, we didn’t really come back with anything there from ASH.


Similarly, a manager at one of the control sites indicated their hope that, if they were part of theintervention, they would have received support on developing a smoking policy:

We don’t have a smoking policy, that was one of the things that I was hoping, if we did have theintervention, that we might have had a bit of, you know, a bit of input with that, that’s one of thethings I was hoping, and that’s why, you know, one of the other reasons why I thought it might beuseful to, to get involved.

Middledale College, FE manager

RESULTS


38

Knowledge of institutional policies was good among staff and students. The intervention sixth form waspart of a school that had a site no smoking policy and, in both the intervention colleges, there were alreadyestablished policies permitting smoking only in designated smoking areas on campus. Both control collegeshad the same rule, although it was only formulated as written policy in one college. In all intervention sites,policies treated e-cigarettes in the same way as tobacco. One control college did not have any institutionalrules about e-cigarettes and the other had recently added e-cigarettes to the smoking policy requiringstudents to vape in designated smoking areas after recognising the issue around the campus. (However, theFE manager indicated that this change was unrelated to their participation in the Filter project.) There wasevidence that, despite being considered the same as tobacco in policies, the products are considereddifferently by students, who may perceive that it is acceptable to use them indoors as an alternative tosmoking outside. As one school staff member said:

I have caught a few. Like if I go in someone’s classroom, like lunchtimes and stuff, like ‘what are youdoing?’, ‘Oh, sorry miss’. So they do try and use them, saves them going outside.


The college policies of smoking only in designated areas were generally well accepted among staff andstudents, and attitudes to areas demonstrated the ubiquity of anti-smoking norms (see Attitudes towardssmoking in the further education context), with some students commenting on the unsightliness andsmelliness of the shelters. On the whole, college staff perceived that having a designated area was apragmatic solution because, whether or not there was provision for them to smoke, ‘they’d findsomewhere. So either it’s controlled to an extent’ (Laurelton College, staff focus group). Although staffand students perceived that people mostly adhered to the smoking policy, they also noted that manystudents smoked just outside the campus boundaries. One student described how ‘they’ll be like outsidethe gates, which is quite annoying when you’re walking past and that’s what you can smell’ (ValesideCollege, student focus group). A staff member at another college described how:

You see the buses arriving in the morning, and they don’t walk to the smoking shelter, they’re lightingup as they’re getting off the bus and smoking, walking towards the smoking [shelter], even thoughthey’re not supposed to do it.


In contrast to the general adherence to, and acceptance of, smoking only in designated areas in the colleges,the sixth form policy of prohibiting smoking on site was not well accepted by students. As one studentexplained, the prohibition of smoking for sixth formers left students who wish to smoke in a difficultposition: ‘we’re not allowed out of school but then we’re not allowed to smoke in school, so where are wesupposed to go?’ (Athervale School, student focus group). Students perceived that the school policy did notcorrespond with their desire for autonomy (see Attitudes towards smoking in the further education context)and liked the idea of designated smoking areas, saying that: ‘Like up the college they’ve got a bus stopwhich is a smoking area [. . .] That’s the thing I like about college’ (Athervale School, student focus group).Staff at the colleges were aware of this and recounted how school students were attracted to collegebecause of the more liberal approach, with the smoking policy being a manifestation of this increasedautonomy:

We used to take school groups around, they’d come for a visit to the college, and, so we’d showthem round outside, and that used to be one of the first questions a schoolchild would ask, are youallowed to smoke in college? [laughter] [. . .] because it’s, we don’t have uniform, they, it was afreer environment.


School sixth form students who wished to smoke did so wherever they could without getting caught. Onenon-smoking student smoke said students smoke ‘wherever there’s no cameras’, with another qualifying,‘or teachers’ (Athervale School, student focus group). It was widely known among school sixth form



39

students that staff have ‘like a rota like, certain like, you certain time like an hour where different teacherswill go outside and like chase up the smokers’ (Athervale School, student focus group). School sixth formstudents explained how staff pursue them up the rugby field to move them away from school buildings.When staff were not on duty, students smoked wherever they liked, as one student described: ‘if they’renot on duty, we’ll just sit on the steps’ (Athervale School, student focus group). The school staff describedtheir response to finding sixth form students smoking and their discomfort in addressing it, saying that:

You’re not going to stop them, so I think it’s better that they’re in sort of one area, it’s not really adesignated, because we don’t have smoking, but, technically we do. And I think we have toacknowledge that we do, and, you know, it, it’s just, a consistent problem and it’s these hardenedsmokers really, isn’t it?


Staff and students recognised that informal smoking areas exist on the school campus, despite anestablished system of sanctions for smoking (letters sent home to parents, followed by exclusion).However, students and staff all said they did not know anyone who had been excluded from school forsmoking, ‘even though it’s school policy’ (Athervale School, staff focus group). Staff and students at theschool sixth form stated that the legal status of sixth form students was different from that of youngerstudents, meaning they were sanctioned differently. One (non-smoking) student explained: ‘The people inour year and the year above, though, the teachers are more laid back about it because, we’re old enough’,with their fellow student adding, ‘And we don’t legally have to be here either, so, that’s why they’re notso concerned about us’ (Athervale School, student focus group). This differed among some students whosmoked, who suggested that they are concerned about being caught smoking because, unlike youngerstudents for whom school is compulsory, ‘They could just kick us out for good, so I think the youngeryears are not that like worried, but us, yeah’ (Athervale School, student focus group).

In contrast, college students were unaware of the sanctions that might be applied if they were caughtsmoking outside the designated areas. Although the FE manager at one control site indicated that ‘there isstaff on monitor because our students are quite sneaky [laughs] they will smoke in inappropriate areas’(Glynbel College, FE manager), staff at the intervention colleges reported fewer problems with enforcingthe designated smoking area policy. Staff at one college explained that the introduction of this policy hadmade a big difference:

I think we used to have a bigger problem last year, where they were smoking just about everywhere,and you know, it was part of our job to tell them, not here, but yeah, we don’t see them smoking inthe wrong places so much these days, not in my experience anyway.


At all intervention sites there were a few staff who smoked and it was evident from observations andfocus groups that, at college, staff used the designated smoking areas although some reported smoking intheir cars to avoid using the shelters because they felt that ‘they don’t want to be mingling with studentswhen they’re having a quiet cigarette. They don’t want to socially mix with them’ (Laurelton College, stafffocus group). Similarly, school sixth form students were aware that staff who smoked went off campus intheir cars to smoke, even though this was forbidden for students. One FE manager suggested that it wasnot appropriate for staff to use designated smoking areas because:

Teachers should be like role models I think. You know and be helping to advocate stop smoking andthings like that, but yet they are out there with the students.

Glynbel College, FE manager

Reflecting their sensitivity to the denormalisation of smoking (see Attitudes towards smoking in the furthereducation context), staff and students were aware that smoking outside campus could detrimentally impactthe institution’s reputation. One staff member from the small college indicated that students smoking just

RESULTS


40

off premises ‘would look quite bad for the college’ and that this had influenced a recent decision not tomove to a no smoking campus (Laurelton College, staff focus group), instead maintaining designatedshelters on campus where smoking was less visible to the wider community. As one school studentexplained, concern to keep student smoking out of the public gaze (students were aware that the schooldeclares itself to be a no smoking school) puts staff in a difficult position when enforcing the nosmoking policy:

They try and like get you off the school grounds, [but] they don’t like people smoking at the shopbecause it still looks bad like a bunch of school girls like smoking at the top.


Staff invoked their professional duty of care to students when discussing the pros and cons of theirinstitution’s smoking policy. For example, staff discussed how the college had considered implementing ano smoking policy across all campuses but did not because ‘they thought the students would then go ontothe main road’ (Laurelton College, staff focus group), which would not be safe. Similarly, school sixth formstaff justified their attempts to contain sixth form students smoking in informal smoking areas, saying:

We have to usher them [students], sort of, to a certain area, but we can’t send them off the premises,because obviously they’re under our care aren’t they? So it becomes a bit of a problem.


Few staff or students expressed opposition to designated smoking areas, which seemed to help limitsmoking on college campuses by reinforcing the denormalisation of smoking. However, the location of thedesignated smoking area was important in terms of whether or not it was used, by whom, and howacceptable it was perceived to be by smokers and non-smokers alike. For example, the FE manager at oneof the control colleges described how the smoking shelter was near the campus crèche and that parentswere concerned about this (Glynbel College, FE manager). Moreover, some staff and students felt thatthey may generate unintended consequences by providing new social spaces and areas where smoking isnormalised. For example, some students who did not smoke indicated that they sometimes go to thesmoking shelter to be with their friends and were equivocal about if this led to them smoking. However,students who smoked suggested that it might encourage non-smokers, saying:

They’re taking in second-hand smoke aren’t they? So they’re also getting some sort of nicotine in theirsystem, at the same time.


Similarly, staff felt that forcing students to use the smoking shelters might normalise smoking and leadthem to increase consumption. As one staff member described:

What I’ve noticed is, for example, the students before they came to college were having probably oneor two fags a day, because they’re mixing with smokers who probably go more regularly, they startpicking up the more regular group, so, because it’s like, they go to smoking shelters more often.When they’re in the smoking shelters, oh do you know what, I can smell your cigarette, give me acigarette please. So, it, they probably, start smoking more.


Staff indicated that the social aspect of designated smoking areas could be problematic for students inother ways, through forcing groups of students together who might not otherwise mix. As one explained:

The less confident students and they’re being put into a mixed, especially the middle shelter there,there could be 50, 60 students there. And they’re starting to feel anxious about going to the smoking



41

shelter, so what happens then is they stand off themselves in a small little place, or their own littlehuddle, and then they’re having, you know, the tellings off, the causes for concern, for not using thesmoking shelter, but they get more anxious about using the smoking shelter as well.


Indeed, some students indicated avoidance of smoking in the designated areas because they did not wantto associate with the groups there, moving to locations just off campus instead. As one group of studentsdescribed, they smoke in the park because ‘it’s the closest’ for them and, even though there is a shelternearby, ‘it’s always full of like [. . .] irritating people’, who are ‘childish’ (Valeside College, studentfocus group).

Staff trainingThe premise of staff training was to support staff in talking to students about smoking. The content ofeach session was determined by staff interest and focused on e-cigarettes, as this was deemed a priorityfor staff at each intervention setting. There was no specific training on illegal sales of tobacco, althoughthis did come up in discussion. Interviews with FE managers suggest that the aim was not clearlyunderstood, as one manager indicated:

What’s the aim of the staff training then? So they get trained on, yesterday it was the e-cig[arette]s. Is itfor them then to go out to students and talk to them and deliver something on that as well then, or?


The same intervention team member delivered all staff sessions, with content broadly the same in eachsession, although with adaptations according to the numbers in each group (most of the staff trainingsessions were smaller than the facilitator had anticipated).

Staff who were trained found the sessions very informative and the training acceptable. They particularlyenjoyed learning in an interactive way on an issue that they felt they had a knowledge gap. As one FEmanager summarised:

The training session was really, really useful, and thoroughly of value and, you know, the heads ofyear still talked about that, and what they’ve learnt from, particularly about e-cigarettes [. . .] theywere thoroughly, enjoyed the way that it was delivered. You know and it was very interactive andhands on, and example, a lot of up to date, factual knowledge, um, and it was also done in a way,you know, that they could test their own knowledge.


However, despite being well received, attendance at the sessions was low, with all sites experiencingdifficulties in releasing staff (especially teaching staff) for training. Staff who attended the training wereoften those in support roles, as they had more flexible timetables than those with direct teaching orsupervisory roles. As the manager at the smaller college noted:

I think we had about 10 in there in the end, about nine or 10, which is a good group and I mean thetraining was good, don’t get me wrong, but we did struggle to be able to release those 10 staff.


The trainer noted that this is a common issue: ‘It’s not just the FE, it’s generally when we do training.It can be a problem get, releasing staff to come along to training’ (ID team member 1), further remarkingthat, in some sites, she was allocated only a 1.5-hour slot for the training when the session would ideallybe 2 hours long.

RESULTS


42

Although the training was well received and perceived as increasing knowledge, it is unclear whether ornot staff expected it to facilitate talking to students about smoking. As one staff member said:

The training is definitely useful, you know, but it’s more for us about kind of signposting to theprofessional people, but we do need some knowledge on it.


The FE manager at the same college explained anticipating that staff would use the training ‘if a studentcame in with a particular query on an e-cigarette or a concern, we use that knowledge then’ (LaureltonCollege, FE manager). However, it was clear that although they found it interesting, staff felt that therewere still a lot of unknown issues around e-cigarettes:

I’ve got to be honest, I thought the e-cig[arette] was, in some ways interesting, to know a little bitmore about it, but the message was quite vague because we don’t know enough about it.


Similarly, in discussions about illegal sales of tobacco, staff felt that it was not their place to question studentsabout it, one saying: ‘I don’t think it’s our, our place for us to question’, and another explaining that:

If you see a student who’s got illegal tobacco, which you would deem to be safe, I’d think, oh fairplay, you’re students, you’re getting it cheaper. The same as you wearing a fake t-shirt and carrying afake handbag. You know, what would you do? Do you report them to the police? I wouldn’t.


Students also questioned the appropriateness of staff talking to them about smoking, reflecting theprevailing culture in FE of fostering self-determination and autonomy (see Intervention feasibility andacceptability). Students indicated that it might seem patronising or invasive:

They’re all, like, they’re older aren’t they, and like most of the people here are over 18 so, like it’s justtelling a grown up not to do something. It’s a bit weird isn’t it?


I don’t see the point, you know, because if you smoke then they know you smoke already, and thenthey’re going to come in try and start lecturing you on smoking, and they already know you smoke,then you’re going to get wound up and argue back and that.


Staff reported that, as well as reminding students to smoke in designated areas (in sites that have suchpolicies), they tend to talk opportunistically to students about smoking but are wary of ‘preaching’ to themabout smoking and encouraging them to quit. They suggested that such messages might be more effectivecoming from another young person (youth worker or another student). As one staff member suggested:

They’re used to us just like telling ‘em what to do, you know, reinforcing different things, I mean if itcomes from somebody else, it’d be a bit more strong, I suppose. There’s only so much informationthey can take from a tutor, I suppose . . .


Another staff member explained how staff are concerned to develop and retain trust with students, withconcerns that talking about smoking might threaten this, and suggesting that staff perceive smoking as aless-serious risk to students than other issues that may present:

When we’re dealing with students and it’s confidentiality, because that’s a big thing with us, becauseif, if they don’t get that, that bond with us, they’re not going to interact with us, or trust us, so you



43

know, it’s, I’ve had more than one to say that I didn’t want his or her parents knowing that theysmoke. I have to respect that, otherwise I’ve lost that trust, so they won’t come back to you, to ironout anything else, so rightly or wrongly, we, we sort of, don’t mention it to parents.


Students corroborated that staff take the chance to speak informally with them about smoking and didnot seem hostile to this, appreciating that the staff were trying to be helpful. As one student described:

There’s a kid in my class who constantly has to have a fag all the time, every break, doesn’t he,[name]? Every break. So the teachers joke about it, but they never like, obviously, because it’s more ofa friendly thing, they’re not, because this is obviously to try and promote them saying anti-smoking.They do say, they don’t like smoking. [. . .] They do make an effort to make that known that it’sStoptober and there is help and everything.


Despite these concerns, staff indicated little time for pastoral work with students and that, althoughtutorials might seem like a potential opportunity to intervene, they actually presented little opportunity forany discretionary activities. Staff at one college whose role is to provide support to students felt thatteaching staff might be resistant to training because of existing demands on their time, saying:

If you said to staff, right, you’ve got to have this training, because you’ve got to deliver this in yourpersonal tutorial, that’s not going to go down very well is it?


Youth workAs the content of the youth work activities was agreed by the intervention delivery team with each site,the content of sessions varied. Two sessions were delivered: one focused on developing a ‘no smoking’campaign and the other looked at the role of advertising in promoting smoking. Neither focused specificallyon illegal tobacco supply or use. Activities were delivered by two different facilitators and the same facilitatordid not do all the sessions at each site. The quality and style of facilitation between sessions was observed todiffer, with observers noting that one of the two facilitators provided little guidance or feedback on the tasksand did not seem to try to engage those students who were not actively participating. The other facilitatorhad difficulty maintaining the interest of the participants in one of the two sessions that they were observedto lead (the smoke-free campaign session), but was more successful in generating rapport with students inthe other session. The different experiences of facilitators may also reflect participants’ lack of interest in thecontent of the sessions on smoke free campaigns.

Students felt that youth work sessions would not change their behaviour and that there were issues aboutwho the sessions were targeted at, and why. Both students and staff felt that it was difficult to identify atarget audience for which youth work sessions would be relevant and appropriate, as reflected in theobservations of these sessions. Participants did not seem engaged in the two sessions in which theydeveloped a smoke-free message campaign and were heard to make comments such as, ‘I don’t evenwant to be here’ and ‘Can we go now?’, with the facilitator remarking afterwards that they felt that theparticipants did not want to be there. However, moments of better engagement were reported, forexample during a discussion of the risks of passive smoking and the session on perceptions of smokingover time, with students appearing genuinely interested by some of the photographs, videos and advertsthe facilitator used to stimulate discussion. They responded well to the facilitator’s questions and, with alittle encouragement, most were willing to talk as a whole group.

Some students found the youth work repetitive and felt that sessions would be more effective if deliveredto younger students (e.g. those in year 9). As one explained, ‘sometimes they learn, they teach us stuffthat we already know’ (Athervale School, student focus group). Students who identified as non-smokersfelt that the sessions were not relevant to them, whereas smokers’ responses ranged from being quite

RESULTS


44

hostile, because they felt that smoking is their choice, to interested but not affected by the content.As one group of students indicated:

Student 1: It was quite effective. It did, does make you think a bit, but unless you’re actually thinkingabout quitting, before doing the training, then I suppose, it won’t really make you quit.

Student 2: Yeah, we went out straight away for a fag didn’t we?

Student 1: Yeah. [laughs] As soon as the training finished. We all went out for a fag!Laurelton College, student focus group

Similarly, a staff member indicated that they were not sure of how generic sessions to both smokers andnon-smokers would work, for staff or students. They said:

If you go into, what is it, year 6, primary school, right. There’s a, there’s a chance that not many ofthem will smoke, so you’re talking mainly to a group of non-smokers so you’re catching them early.If you go into tutorial session with us, you could have, you can have smokers and non-smokers.And it would be the same in staff training. In the training we were in we were all non-smokers, sowe’re all kind of nodding our heads and agreeing, if there was a smoker in there, he’d probably bequite defensive and would be giving bigger arguments. And it would be the same in a tutorial session.You’re going to have the non-smokers who would go, yeah, agree, and then you’ve got thenon-smokers who would just be challenging.


Unlike the staff training, which did not require much adaption for FE settings, The Filter team made somechanges to their youth work sessions in anticipation of the slightly older age group in FE. As one memberof the intervention said, ‘[the sessions are] still all interactive that they can get involved with but it wasmore like, to try and get discussion as well.’ They went on to contrast their experience in schools with theirrecent experiences in FE, describing how:

In an FE setting, if there’s less young people then it can . . . they can be, like steer the session morethemselves, like go with like what they’re interested in rather than just giving them activities thatthey don’t want to do anyway. So like, a lot of them in [Valeside College] talked about e-cigarettes.So then it was more like tailored towards that.

Intervention team member 2

The youth worker who delivered the sessions at the small college felt that the activities based arounddeveloping a smoking campaign ‘got a bit stagnant towards the end’, describing how ‘It was a new sessionso it was, you know, you try it out – sometimes it works, sometimes it doesn’t’ (intervention team member 4).

However, the flexibility of the sessions was problematic for FE managers when trying to recruit staff toengage their students with the sessions. As the manager at the large college noted, ‘I think the, the aim ofit, of the session was a little bit woolly I think’, although they were able to describe the content of thesessions, saying:

The sessions were about marketing wasn’t it? And it was about young people’s attitudes towardssmoking and what it influences them regarding, is it peer pressure, is it what they see in the papers oron television, that kind of thing.


The sixth form youth work activities consisted of three consecutive sessions with the same class ofstudents, whereas in the colleges there were one-off sessions with different groups, consisting of betweenfive and 15 students. There were issues of continuity of attendance in the school sixth form, with the



45

youth worker who ran sessions in both the school sixth form and the large college suggesting that theymight have structured the sessions differently if they had multiple opportunities to engage with the samegroup of students. The youth worker felt that the sessions in the school sixth forms had been more‘challenging’ (intervention team member 2) compared with those in the large college, where students hadseemed to be more engaged. They speculated as to the reasons for this, saying it might be because therewas a smaller group in the college and also:

I think [it was] ‘cos it was their PSE [personal, social education] lesson and I think they’d been told tocome for a smoking workshop, so that doesn’t really go down well anyway [laughs] and I think a lotof them had, I don’t know whether it was the time of the year, a lot of them wanted to do like coursework and but the teacher made them come in to the session. But then I guess it was different to theirusual classes in the college as well, so don’t know, it’s hard to tell really.

She went on to add that the students ‘seemed a lot more mature in [Valeside] college as well [laughs]’.The other youth worker also delivered sessions in the large college (n = 5) and similarly felt that thesessions had been well received by students. In contrast to the facilitator’s impression, though, the FEmanager at the school sixth form felt the training had:

Made them [students] think about whether they were going to carry on smoking, it was certainlyraising their awareness with them. Whether they had actually any impact and they stopped smoking,I really couldn’t say.


In the large college there was a teacher present who participated and the youth worker felt that thiswould have also helped the session run more smoothly in the school sixth form.

One control FE manager perceived their participation in the Filter research project had highlighted the issueof student smoking with staff, resulting in additional lessons ‘to raise awareness or just discussions anddebates with students in the classroom’ (Glynbel College, FE manager).

Social mediaIt appeared that FE settings were reluctant to give smoking prevention messages prominence online, andat one of the sites the website was not adequate or active enough to be used. The intervention deliveryteam member responsible for this component of the intervention sent one tweet and a Facebook messageto each intervention site each week, but very few posts were retweeted or shared without prompting.They stated:

With Facebook there just wasn’t the engagement there. Um, on Twitter it was a lot better, um,so I think there was probably about 12 posts again, something around that, and the majority of theFEs did retweet the ones that I directly tweeted them.

Intervention team member

Most students said that, although they access social media (e.g. Twitter, Facebook, Instagram), theydo not really engage with messages from third parties on these forums and tend to ignore unsolicited/non-personalised messages unless they are already interested in what is being advertised. As one studentindicated:

I just don’t think people would really pay attention to it [information]. I think there’s more interestingthings you’d rather like I think.


RESULTS


46

Another student from the same college explained:

It’s got to be your choice that is, some people will just go straight past, whereas if you’ve got it in yourhead that you want to quit you’re going to think, oh yeah, I’ll click on that and have a look, so. Itdepends on the person really.


Students reported gathering a lot of information from social media but that it needed to be carefullytargeted and age appropriate. They indicated that unwelcome messages could make them ‘switch off’completely. As one staff member said, this was similar to students feeling badgered by their parents:

Staff 1: If you’re using things like Twitter, it’s got to be the right message and probably not too often,because I, I’ve got various Twitter accounts and there’s certain accounts that keep on popping up andyou just, do you know what, I’m bored. So you just unfollow. So if the message is coming constantly,and you’re bombarding these messages, just click unfollow and then you’ve lost them straightaway.So it needs to be relevant and . . .

Staff 2: Just desensitises you again.

Staff 3: Yeah.

Staff 1: It’s just like, oh, whatever, whatever, isn’t it? You know.

Staff 2: The more you tell, the more parents tell teenagers what to do.

Staff 3: Yeah.

Staff 2: The less likely they are to do it anyway.Laurelton College, staff focus group

Similarly, FE managers were cautious about inundating students with messages, while also feeling that theintervention messages should have been more obvious around campus. As one said: ‘people are justinundated with information and it’s just too much, you’ve got to be careful as to what you do get across’(Valeside College, FE manager). The FE manager at one intervention site indicated they had retweetedmessages from Filter and ASH Wales a couple of times, and some information had been included on theintranet, but they had not engaged much with social media. Students had very little awareness of anysocial media messages from ASH Wales and stated that they rarely, if ever, access the school or collegewebsite. Those who had seen something online had seen it in the youth work session.

In one discussion among college staff, they indicated that they thought social media messages needed tobe carefully applied, as it might prompt students to smoke more. They speculated that:

It’s done on social media and it’s flashed at them, it makes them think they want a cigarette then? Itsorts of brings it to their attention. If you don’t bring it to their attention, they’re fine getting on withthings. The moment you bring it to their attention, it’s like ‘Right, I want one now’. So that can have abit of an adverse effect.


Perceived impact and awareness of the interventionOverall, the intervention had limited reach to staff and students. Very low levels of awareness of The Filteryouth project and ASH Wales among staff and students reflect the weaknesses and challenges withimplementation described above.



47

As the manager at Valeside College commented, ‘I don’t think they would have been aware of the project,because we, you know, we haven’t been, pushing it, pushing it, pushing it’. A few students in theintervention school sixth form had heard of The Filter project through having received other outreachyouth work the previous year. FE managers could not recall all the intervention components, with onesaying: ‘I’m going to struggle to answer questions about the intervention’ (Laurelton College, FE manager).

Some students participating in focus groups could recall participating in youth work sessions, and staffrecalled receiving training. All students and staff were already aware of their site smoking policy, reflectingawareness of changing social norms around smoking and non-smoking normalised in practice (see Attitudestowards smoking in the further education context). As one student put it:

Obviously [I’ve] seen the signs and that obviously that’s a normal thing. And obviously that’s ourgeneration, and I’m sure back in the day they wouldn’t have had those signs. Um, there are smoke-free signs in college isn’t there? That say no smoking?


Staff and students seemed most engaged when discussing issues relating to the site policy on tobacco,although there was little input from the intervention delivery team on this. Staff who received trainingseemed to enjoy the sessions and found them informative. Student reactions to the youth work weremixed, although staff felt that they might have appreciated the sessions ‘because they do enjoy that kindof tutorial stuff. It gets them out of their own boring lessons’. However, they doubted the impact of thesessions, saying: ‘I don’t think it would have put anyone off smoking’ (Laurelton College, FE manager).Another FE manager concurred, stating that:

The youth work sessions, you know. With, without a doubt, um, you know, I think are really, reallyimportant [. . .] in an ideal world I would have liked to have done more youth work sessions. [. . .] justto get the message, just to get, just to get more students, and make them aware of the project, andmake them aware of what you’re trying to achieve.


The core message of the intervention (smoking prevention) was not clearly communicated; some studentsassumed that the intervention was ‘anti-smoking’ and smokers were not receptive to this. Similarly,students who were non-smokers felt that it was not necessary for them to be bombarded with thismessage. Students and staff felt that smoking prevention messages might be too late for this populationgroup, but that improved smoking cessation support could be helpful. As one FE manager summarised:

To be honest, if by 16 they haven’t started yet, I don’t, I know they come to college then, I supposethey see, they may get new friends who do smoke, but they’re probably going to be trying it a lotyounger than 16. Um, but maybe to prevent smoking but to also assist with quitting as well.


SummaryThis section has described in detail how each of the intervention components was received by staff andstudents. It shows that, despite a willingness to engage in an intervention in FE, including via policychange at the institutional level (as reported in Barriers to acceptability and implementation), the reach andimpact on students and staff in terms of awareness of the intervention was very limited. In the light of theanalyses presented in Intervention feasibility and acceptability, which indicate that smoking is largelydenormalised among the FE population and that students and staff value freedom, personal responsibilityand self-determination, it is perhaps unsurprising that core components of this intervention, such as stafftraining and youth working, were not perceived as relevant and necessary in the FE setting.

RESULTS


48

Research methods: feasibility and acceptabilityThis section focuses on RQ 10: how acceptable were the data collection methods to students and staff,and do participants think longer-term follow-up via e-mail or telephone interview would be feasible? First,we present FE managers’ perceptions of the randomisation and recruitment processes. Then we describethe acceptability and feasibility of the survey data collection methods used at baseline and follow-up in thispilot study.

Randomisation and recruitmentFurther education managers stated that they were satisfied with the trial recruitment process (most wereapproached by e-mail by the intervention delivery team). Agreement to participate had usually beencontingent on senior management approval, although one manager noted that it had been delegated tothem, saying:

This just sort of fell into my lap a bit, but it’s not, it’s not, you know, I guess I was the nearest personto, you know, to sort of pick it up.


In the smaller control college, students were involved in the decision regarding whether or not toparticipate, and according to the FE manager they had ‘a good input from the students, they were allexcited about it’ (Glynbel College, FE manager).

Further education managers perceived the randomisation process as fair and acceptable. Some expresseddisappointment that they were not in the intervention arm, but said that this possibility did not affect theirdecision to join the trial. A manager from a control site summarised the tenor of managers’ feelings aboutrandomisation, saying:

Our students could only benefit and if they, if we weren’t chosen then things sort of stayed more or less,that’s what I thought at the time anyway, I thought that, you know, there was no harm done then.


Further education managers had few reservations about getting involved in the study, although some hadbeen concerned about the time commitment required. One manager described their reservations aboutwhich campus of the college was selected for randomisation and, after being asked to suggest a smallcampus, stated that they ‘didn’t think that we chose the right campus in all honesty’ because they knewthat there were not many smokers on the small campus and also because the campus was located in avery rural area. There were no shops ‘within a 3-mile radius’, which they felt would therefore renderredundant any intervention with shopkeepers about access to tobacco (Laurelton College, FE manager).

Further education managers reported several motivations for participating in the study. Primarily, managerswere keen to work with other organisations, and for some it was part of their role to foster links withexternal agencies. One manager suggested that the reputation of these organisations was important:

When you got professional bodies like the Filter team and ASH, if they can come in and offersomething different, we’ll always take that.


The same manager indicated that they had felt that it would be good for students to see the university atwork, saying:

We are happy to support the university, it was good to work with the uni[versity] and us. You know,it created awareness of the university as well for our students.




49

Managers were also motivated by potential benefits for students, with most perceiving that smoking wasan issue among their students:

My interest was in, you know, providing support um, and information to our staff and students aboutsmoking. Because it is an issue.


Another reason for participating in the study was that they might learn as an organisation. As onemanager explained, ‘I thought it would benefit the students in the school, and that we might get somefeedback from it that would allow us to, um, further support the students’ (Athervale School, FE manager).One of the managers at a control site indicated that they were disappointed not to be involved becausethey thought it might help them to develop their site policy on smoking.

Acceptability of the survey data collection methodsOn the whole, surveys were an acceptable form of data collection and, although longer-term follow-up bytelephone or e-mail was perceived as acceptable, both students and staff felt that it was unlikely to befeasible. However, there was some indication of fatigue among students about completing surveys,particularly on smoking. For example, one student said that:

People ain’t going to be willing because the smokers then probably find it offensive, and then they’renot willing to help people like yourselves like this then are they? Because they think ah, they’re justtrying to offend us. So why should we help.


There was no strong preference among students for completing the surveys online or on paper, althoughonline (including via mobile phone) was perceived as most convenient.

Accessing students to complete surveys was challenging. FE managers had concerns about data protectionand it was not possible to share student e-mail addresses so that the surveys could be sent directly tothem, so students were asked to complete the survey at freshers’ fairs. This was acceptable to bothstudents and staff, although, as one FE manager put it, ‘The timing probably wasn’t brilliant for us becauseSeptember/October is people really settling into their courses’ (Laurelton College, FE manager). Moreover,this system made it harder to trace students for the follow-up survey as they were approached at random.Indeed, after an initial registration period at the start of each academic year, students may move todifferent campuses or courses, or may attend the site only sporadically or on particular days (e.g. they mayhave started a work placement). As one FE manager explained:

Our learners tend, full-time learners tend to be in 3 days a week, and they could be any 3 days of 5,so, you know, come in on a Monday, not everybody’s going to be in on a Monday, not everybody, doyou see what I mean? So it’s from that sort of point of view, it’s a very flexible provision, and um, andtherefore you may not always get, you know, the responses, and it might, it just might mean thatresearchers need to come a bit more often to those sort of organisations or campuses really.


For the follow-up survey, students were approached in public spaces on campuses. A few students indicatedthat they did not like being disturbed on their breaks to complete the follow-up survey (e.g. ValesideCollege, student focus group). Staff and students suggested that students might be asked to complete thesurvey in their tutor groups, although students indicated that they might then feel obliged to complete it.For example, one sixth form student said: ‘if you’re told to do it in a lesson like fill out this questionnaireplease, you will’ (Athervale School, student focus group). However, students did not seem to mind this,saying: ‘It’s less of a chore then do you know what I mean?’ (Valeside College, student focus group).

RESULTS


50

Students and staff felt using personal e-mails would be the best way to follow-up students, but allacknowledged that few students check e-mails. Many students said using telephone or text message tocontact them would be acceptable. For example, one student said:

I’m fine with anything. I’m easy, but obviously you’ve got to think as well. Phone numbers mightchange in a year or 2. Because my phone number’s changed three times last year, because I wasupgrading my contract.


However, one FE manager indicated that even now, when they try to contact students via telephone, ‘theywon’t answer if it’s a withheld [number]’ (Glynbel College, FE manager). Another FE manager reiteratedthe difficulty in following up with students over time, stating that:

To be honest at the start at the year they were all very keen and they want to do that, but doing thefollow-up and doing the evaluation they were less keen.


Students had mixed views on the use of incentives (win an iPad, ‘Love2shop’ vouchers) to encourage themto complete the survey. Although they did not believe it undermined the credibility of the survey, they haddiffering views about whether or not it would be an effective inducement. One student said, ‘it was thatthey offered prizes, that’s the only reason I did it really’ (Laurelton College, student focus group). Staffalso seemed to think that incentives would be helpful, as one staff member said, ‘They’re not going to doit for nothing’ (Laurelton College, staff focus group). On the other hand, some students felt that if theresearcher was friendly, introduced themselves and made you feel comfortable, then ‘that would be fine’(Valeside College, student focus group). A student at a different site indicated that:

I don’t think you should offer anything. To be quite honest. It’s like, if they want to fill it out, they will,and then if they don’t, then don’t it.


The shopping vouchers, in particular, were seen as acceptable and appropriate incentives. Some studentsfelt that the prize of an iPad was too valuable for them to have a realistic chance of winning it and thatmore, smaller, prizes such as shopping vouchers, would be more effective. Students were cynical abouttheir chances of winning prizes, comparing it with incentives they see on television and online. They feltthat incentives did not motivate them to participate because:

Everybody thinks it’s a fix though when people, say like oh, take this survey and you’re going to get aprize. Everyone thinks it’s a fix, so you don’t actually think you’re going to win.


Students needed to feel that they had a chance of winning for incentives to work, and were scepticalabout the incentives because they had not heard who had won the iPad at baseline. One group ofstudents suggested that it should be clear that there would be one prize per site so that they would knowwho had won. Alternative incentives suggested included smoking cessation advice: ‘free help stoppingstuff’ (Valeside College, student focus group).

SummaryOverall, the FE managers were supportive of the research, including the use of randomisation, and theywere keen to engage with external agencies to deliver interventions to students. However, FE settings arenot often engaged in trials and may therefore require additional support and information to ensure thatthe research process is clear. Digital survey methods and use of incentives are likely to be the mosteffective methods for surveying students at multiple time points in this context, but more methodologicalwork is required to understand how to work with organisational settings in which barriers to identificationof appropriate sampling frames exist. More effective approaches to initial organisational process mappingwould improve recruitment and retention rates if further research were to be conducted in this setting.



51

Pilot primary and secondary outcomes

The third objective of this study was to pilot primary, secondary and intermediate outcome measures andeconomic evaluation methods and measures. It was not an objective of the pilot study to assess interventioneffects and the study was not designed or powered to do so. The response rates, prevalence and distributionby arm are reported for the pilot primary and secondary outcomes at baseline (see Missing data, prevalenceand distribution by arm at baseline) and follow-up (see Missing data, prevalence and distribution by arm atfollow-up). This section concludes by assessing the feasibility of assessing cost-effectiveness using a cost–utilityanalysis based on the EQ-5D-5L (see Feasibility of assessing cost-effectiveness)

Missing data, prevalence and distribution by arm at baselinePilot primary and secondary outcome measures at baseline are reported in Tables 7 (categorical data)and 8 (numerical data), by arm and overall.

Missing dataThere were few missing values with the percentage of missing responses being < 2% for all but fourquestions: living with an adult in paid work, spending money each week, number of cigarettes per dayand the heaviness of smoking (as measured using the HSI). Across these variables there were more missingdata in control than in intervention arm participants. For living with an adult in paid work, the percentageof missing responses in the control arm was 4.1% and in the intervention arm was 1.7%. The percentageof missing data for spending money each week was not equally distributed across arms (control, 11.3%;intervention, 5.9%). For the number of cigarettes per day and the HSI, the percentage of missing data was7.2% in the control arm and 2.5% in the intervention arm.

Prevalence and distribution by armThe indicative primary outcome in this study was the prevalence of weekly smoking, which was 20.6%(n = 233) for all participants (95% CI 18.4% to 23.1%). There was a greater percentage of weekly smokers inthe control arm than in the intervention arm [23.6% (n = 111) vs. 18.5% (n = 122)] (Table 9). This compareswith 20.0% of 16- to 19-year-olds in the 2010 GLS.2 Twenty-one per cent of control arm and 14.9% ofintervention arm participants smoked every day. Around one-quarter of participants in each arm had eversmoked cannabis. Out of those who had ever tried cannabis, 43.1% of control and 40.9% of interventiongroup participants had smoked cannabis in the past 30 days. There were more control group participantswho had smoked cannabis on ≥ 4 of the last 30 days (control, 60.4%; intervention, 48.7%). The majority ofparticipants screened positive for hazardous levels of alcohol consumption using the ≥ 2 cut-off point on theAUDIT-C validated in adolescents45 (overall 68.2%); slightly more screened positive in the intervention groupthan in the control group (70.9% vs. 64.5%). The percentage screening positive using the ≥ 5 cut-off point,validated in adults,38 was evenly distributed across arms (control, 36.4%; intervention, 38.5%).

The ICC suggested there was a moderate level of clustering at baseline in weekly smoking status (ICC 0.03,95% CI 0.00 to 0.08), lifetime smoking status (ICC 0.02, 95% CI 0.00 to 0.10) and screening positive forhazardous levels of alcohol consumption using the ≥ 2 cut-off point on the AUDIT-C (ICC 0.02, 95% CI0.00 to 0.21). The ICCs for the use of cannabis in the last 30 days (ICC 0.12, 95% CI 0.03 to 0.39) andmore than four times in the last 30 days (ICC 0.12, 95% CI 0.02 to 0.54) were larger. In contrast, there wasvery little clustering in lifetime cannabis use (ICC < 0.00001). There was a moderate level of internalconsistency across items in the HSI (α = 0.50) and AUDIT-C (α = 0.65).

The distribution of numeric baseline variables for control and intervention group participants are summarisedin Table 8. The five-number summaries show a wide range of values reported on the AUDIT-C, for thenumber of cigarettes per day and the HSI, with little difference across study arms. Median scores on theAUDIT-C (control, 3.0; intervention, 4.0), number of cigarettes per day (control, 10.0; intervention, 10.0) andHSI (control, 2.0; intervention, 1.0) were very similar across arms.

RESULTS


52

Missing data, prevalence and distribution by arm at follow-upThe characteristics of eligible and non-eligible participants according to the pilot primary and secondaryoutcome measures at follow-up are reported in Tables 9 (categorical data) and 10 (numerical data), by armand overall.

Missing dataAs at baseline, there were few missing values, with the percentage of missing responses for the pilot primaryand secondary outcome item typically only 1–2%. Students who provided data at baseline and follow-uphad a slightly lower prevalence of weekly smoking (16.7% vs. 22.8%; p = 0.03) and lifetime cannabis use(23.5% vs. 29.5%; p = 0.02) than students who provided data only at baseline. No meaningful differencewas found on any demographic characteristic or other secondary outcomes.

TABLE 7 Pilot primary outcome and categorical secondary outcomes at baseline

Variable

Baseline data: distribution over categories by trial arm (%)


Ever tried smoking, even if a puff?

Missing 1.1 1.4 1.2

Yes 54.0 52.3 53.2

Do you smoke at all nowadays?

Missing 1.1 1.4 1.2

Yes, every day 20.6 14.9 17.3

Yes, at least once a week 3.0 3.6 3.4

Yes, occasionally but not once a week 6.8 8.8 8.0

No, never 68.5 71.4 70.2

Weekly smoking status

Missing 1.1 1.4 1.2

Weekly smoker 23.6 18.5 20.6

Ever smoked cannabis?

Missing 1.5 1.8 1.7

Yes 26.2 28.2 27.4

Smoked cannabis in past 30 days?a

Missing 0 0 0

Yes 43.1 40.9 41.8

Smoked cannabis on ≥ 4 days in last 30 days?b

Missing 0 0 0

Yes 60.4 48.7 53.5

AUDIT-C

Missing 1.1 2.0 1.6

Hazardous drinker (scoring ≥ 2)c 64.5 70.9 68.2


a Percentages calculated only in those who reported ever smoking cannabis.b Percentages calculated only in those who reported smoking cannabis in last 30 days.c AUDIT-C: cut-off point validated in adolescents is ≥ 2; and in adults is ≥ 5.



53

TABLE 8 Numerical secondary outcomes at baseline

Variable (potential scale range) Trial arm n

Baseline data: demographic characteristics and outcomes by trial arm (%)

Missing Minimum 25th centile Median 75th centile Maximum Mean SD

AUDIT-C (0–12) Control 465 1.1 0.0 0.0 3.0 6.0 11.0 3.4 2.9

Intervention 647 2.0 0.0 1.0 4.0 6.0 11.0 3.6 2.7

Overall 1112 1.6 0.0 1.0 3.0 6.0 11.0 3.5 2.8

Number of cigarettes per daya (0–100) Control 111 7.2 0.0 5.0 10.0 16.0 50.0 12.1 8.4

Intervention 122 2.5 0.0 5.0 10.0 15.0 35.0 10.5 6.5

Overall 233 4.7 0.0 5.0 10.0 15.0 50.0 10.7 7.6

HSIa (0–6) Control 111 7.2 0.0 0.0 2.0 3.0 5.0 1.8 1.6

Intervention 122 2.5 0.0 0.0 1.0 3.0 5.0 1.6 1.5

Overall 233 4.7 0.0 0.0 2.0 3.0 5.0 1.7 1.5

a Information only recorded on weekly smokers.

RESULTS

NIHRJournals

Librarywww.journalslibrary.nihr.ac.uk

54

Prevalence and distribution by armAcross all participants the prevalence of weekly smoking was 17.2% (95% CI 13.9% to 21.2%).There was imbalance in prevalence, with more weekly smokers in the control arm (21.3%) than in theintervention arm (14.3%). The percentage of participants who smoked every day was slightly higher in thecontrol arm than in the intervention group (control, 16.7%; intervention, 13.0%), as was the proportionsmoking at least once per week (control. 4.6%; intervention, 1.3%).

Around one-quarter of participants had ever smoked cannabis, slightly more in the intervention group(26.1%) than in the control group (21.8%). Of those who had ever tried cannabis, 44.7% of control and

TABLE 9 Pilot primary outcome and categorical secondary outcomes at follow-up

Variable

1-year follow-up data: distribution over categories by trial arm (%)


Ever tried smoking, even if a puff?

Missing 0.6 0.8 0.7

Yes 60.3 55.9 57.7

Do you smoke at all nowadays?

Missing 1.2 0.8 0.9

Yes, every day 16.7 13.0 17.3

Yes, at least once a week 4.6 1.3 3.4

Yes, occasionally but not once a week 6.9 7.6 8.0

No, never 70.7 77.3 70.2

Weekly smoking status

Missing 1.2 0.8 0.9

Weekly smoker 21.3 14.3 17.2

Ever smoked cannabis?

Missing 1.2 0.8 1.0

Yes 21.8 26.1 24.3

Smoked cannabis in the past 30 days?a

Missing 0 0 0

Yes 44.7 32.3 37.0

Smoked cannabis for ≥ 4 days in last 30 days?b

Missing 0 0 0

Yes 88.2 85.0 86.5

AUDIT-C

Missing 0 0 0



a Percentages calculated only in those who reported ever smoking cannabis.b Percentages calculated only in those who reported smoking cannabis in last 30 days.c AUDIT-C: cut-off point validated in adolescents is ≥ 2; and in adults is ≥ 5.



55

TABLE 10 Numerical secondary outcomes at follow-up


1-year follow-up data by trial arm (%)


AUDIT-C (0–12) Control 174 0.0 0.0 1.0 3.5 6.0 11.0 3.6 2.6

Intervention 238 0.0 0.0 2.0 4.0 6.0 10.0 3.7 2.5

Overall 412 0.0 0.0 1.0 4.0 6.0 11.0 3.6 2.6

Number of cigarettes per daya (0–100) Control 37 0.0 0.0 5.0 8.0 13.0 30.0 10.1 7.3

Intervention 34 0.0 0.0 6.2 10.0 15.0 25.0 10.9 6.2

Overall 71 0.0 0.0 5.0 10.0 15.0 30.0 10.5 6.8

HSIa (0–6) Control 37 0.0 0.0 0.0 2.0 3.0 4.0 1.6 1.3

Intervention 34 0.0 0.0 0.0 2.0 3.0 4.0 1.6 1.6

Overall 71 0.0 0.0 0.0 2.0 3.0 5.0 1.6 1.6

a Information only recorded on weekly smokers.

RESULTS

NIHRJournals


56

32.3% of intervention group participants had smoked cannabis in the past 30 days. The percentage ofcontrol and intervention group participants who had smoked cannabis on ≥ 4 of the last 30 days wassimilar (control, 88.2%; intervention, 85.0%). Seventy-five per cent of participants screened positive forhazardous levels of alcohol consumption at follow-up, using the ≥ 2 cut-off point on the AUDIT-C, withslightly more screening positive in the intervention group than in the control group (77.3% vs. 71.3%).The percentage screening positive using the ≥ 5 cut-off point validated in adults was similar in both arms(control, 36.2%; intervention, 37.4%).

The distribution of numeric outcomes for control and intervention group participants is summarised inTable 10. Across groups, the median AUDIT-C score was 4.0, number of cigarettes per week was 10.0 andthe HSI score was 2.0. The five-number summaries suggest that there was a large range of scores on theAUDIT-C (0–11), number of cigarettes (0–30 per day) and HSI (0–5), and all had relatively large SDs. Therewas little variation across the intervention and control arms for medians on the AUDIT-C (control, 3.5;intervention, 4.0) and HSI (control, 2.0; intervention, 2.0), and participants in the intervention arm smokedslightly more cigarettes than those in the control arm (control, 8.0; intervention, 10.0). No formalcomparisons were carried out, so any differences must be interpreted with caution.

There was a high level of clustering in weekly smoking status (ICC 0.08, 95% CI 0.00 to 0.21) and the useof cannabis in the last 30 days (ICC 0.11, 95% CI 0.01 to 0.60) at the 1-year follow-up. In contrast, therewas very little clustering in lifetime smoking status (ICC 0.01, 95% CI 0.00 to 0.19), lifetime cannabis use(ICC < 0.00001) and use of cannabis more than four times in the last 30 days (ICC < 0.00001). The internalconsistency across items in the HSI (α = 0.17) was poor and in the AUDIT-C (α = 0.58) was moderate.

Twenty per cent of students (n = 233) were weekly smokers at baseline. A number of smokers werelost to follow-up, such that baseline prevalence of weekly smoking among those who remained in thestudy at follow-up was 16.8% (n = 69). Of these weekly smokers, 18 (5.3%) were no longer weeklysmokers at follow-up. There was no discernible difference by arm, with 7.4% (n = 10) in the control and4.0% (n = 8) in the intervention arm not classified as a weekly smoker at follow-up. Of the 336 studentswho were not a weekly smoker at baseline and had data at follow-up, 21 (6.3%) were classified as aweekly smoker at follow-up. The difference in weekly smoking uptake was slightly higher in the control(8.9%, n = 12) than in the intervention arm (4.5%, n = 9).

Exploratory multilevel logistic regression models adjusting for baseline weekly smoking status, age, gender,residence with an employed adult, ethnicity and educational attainment (≥ 5 GCSEs at A*–C) indicatedthat there was very little difference in the risk of weekly smoking at follow-up between the interventionand control students (odds ratio 0.93, 95% CI 0.23 to 3.76).

To examine the reliability of reporting on having ever smoked, we calculated the percentage of participantswho recanted. This is when participants provide an illogical permutation of responses. We used responses tothe question, ‘Have you EVER tried smoking a cigarette, even if it was only a puff or two?’. Participants whorecanted were those at baseline who said they had ever smoked and then at the 1-year follow-up said theyhad never smoked. Of the 412 participants, 206 (50.0%) at baseline and 238 (57.7%) at the 1-year follow-uphad ever tried smoking. Out of the baseline ever smokers, 17 (8.3%) recanted by responding that they hadnever smoked at the 1-year follow-up.

Feasibility of assessing cost-effectiveness

EuroQol-5 Dimensions, 5-level version items and health service use at baselineParticipants were requested to tick one box that best describes their health today across five domains:mobility, self-care, usual activities (e.g. work, study, housework, family or leisure activities), pain/discomfortand anxiety/depression. Table 11 shows the distribution of responses to each item at baseline. They were thenasked to put a cross on a line to indicate how good or bad their health is on a scale ranging from 0 (worst



57

TABLE 11 Categorical EQ-5D-5L items and health service use at baseline

Variable



Mobility

Missing 3.6 3.2 3.4

I have no problems in walking about 88.7 91.1 90.1

I have slight problems in walking about 5.1 4.5 4.8

I have moderate problems in walking about 1.5 0.6 1.0

I have severe problems in walking about 0.4 0.2 0.3

I am unable to walk about 0.6 0.5 0.5

Self-care

Missing 3.6 3.5 3.5

I have no problems washing or dressing myself 92.8 94.4 93.7

I have slight problems washing ordressing myself

1.7 1.1 1.3

I have moderate problems washing ordressing myself

0.6 0.7 0.7

I have severe problems washing ordressing myself

0.9 0 0.4

I am unable to wash or dress myself 0.4 0.3 0.4

Usual activities

Missing 3.8 3.8 3.8

I have no problems doing my usual activities 85.3 88.5 87.2

I have slight problems doing my usual activities 7.7 5.2 6.2

I have moderate problems doing myusual activities

2.3 1.8 2.0

I have severe problems doing my usual activities 0.4 0.5 0.4

I am unable to do my usual activities 0.4 0.3 0.4

Pain/discomfort

Missing 4.3 3.5 3.8

I have no pain or discomfort 75.3 72.6 73.7

I have slight pain or discomfort 13.4 18.9 16.6

I have moderate pain or discomfort 4.0 3.9 4.0

I have severe pain or discomfort 2.8 0.8 1.6

I have extreme pain or discomfort 0.2 0.3 0.3

Anxiety/depression

Missing 4.0 3.6 3.8

I am not anxious or depressed 69.8 60.6 64.4

I am slightly anxious or depressed 14.0 18.9 16.9

I am moderately anxious or depressed 6.4 11.7 9.5

I am severely anxious or depressed 3.6 4.1 3.9

I am extremely anxious or depressed 2.1 1.1 1.5

RESULTS


58

health you can image) to 100 (best health you can image). Table 12 shows the distribution of responses foreach numerical item.

Health service use over the last 30 days was defined as going to see a doctor or nurse about a health problem,excluding visits for contraceptive advice. For those who had visited a doctor or nurse, they were asked on howmany occasions. Hospital admittance was also assessed over the last 30 days. Those who reported beingadmitted to hospital were asked whether it was for day care or the number of nights. Table 11 shows thataround one-fifth of participants had visited a doctor and 5% had been admitted to hospital.

EuroQol-5 Dimensions, 5-level version items and health service use at follow-upParticipants were asked to repeat the EQ-5D-5L and health service use questions at follow-up. As atbaseline, they were asked to tick one box that best describes their health on the day of survey overfive domains: mobility, self-care, usual activities (e.g. work, study/housework, family or leisure activities),pain/discomfort and anxiety/depression. Table 13 shows distribution of responses to each item atfollow-up. As with the baseline survey, participants were also asked to put a cross on a line to indicatehow good or bad their health is on a rating scale from 0 (worst health you can imagine) to 100(best health you can imagine). Table 14 shows the distribution of numerical responses and reporteduse of health services at follow-up for those respondents who indicated that they had visited a doctoror nurse over the last 30 days. The control group mean at follow-up was 1.8 visits (SD 1.3 visits), theintervention group mean at follow-up was 1.7 visits (SD 1.1 visits). The means for the EQ-5D-5L indexscore in the control and intervention group were both 0.9 (SD 0.1).

TABLE 11 Categorical EQ-5D-5L items and health service use at baseline (continued )

Variable



During the last 30 days, have you visited a doctoror nurse?

Missing 1.7 2.6 2.2

Yes 17.0 22.6 20.3

During the last 30 days, have you been admittedto hospital?

Missing 2.3 2.6 2.5

Yes 4.3 5.8 5.1

How many nights admitted to hospital in last30 days?a

Missing 0 0 0

Day case 65.0 65.8 65.5

1 25.0 18.4 20.7

2 0 2.6 1.7

≥ 3 10.0 13.2 12.1

a Information only recorded on those who had been admitted to hospital in last 30 days.



59

TABLE 12 Numerical EQ-5D-5L items and health service use at baseline


Baseline data: characteristics by trial arm (%)


How good or bad is your health TODAY?[0 (worst you can imagine) to 100 (best you can image)]

Control 449 4.5 0.0 72.0 85.0 96.0 100.0 80.0 21.5

Intervention 639 3.2 6.0 75.0 87.0 95.0 100.0 81.8 17.5

Overall 1081 4.4 0.0 74.0 87.0 95.0 100.0 81.1 19.3

How many visits to a doctor or nurse in last 30 days? (0–6)a Control 78 2.5 1.0 1.0 2.0 3.0 5.0 1.9 1.1

Intervention 148 0.7 1.0 1.0 1.0 2.0 6.0 1.6 1.0

Overall 226 1.3 1.0 1.0 1.0 2.0 6.0 1.7 1.0

EQ-5D-5L index score (–1.0 to 1.0) Control 470 0.0 –0.3 0.9 1.0 1.0 1.0 0.9 0.1

Intervention 660 0.0 0.1 0.9 1.0 1.0 1.0 0.9 0.1

Overall 1130 0.0 –0.3 0.9 1.0 1.0 1.0 0.9 0.1

a Information only recorded for those who had visited a doctor or nurse in last 30 days.

RESULTS

NIHRJournals


60

TABLE 13 Categorical EQ-5D-5L items and health service use at follow-up

Variable

1-year follow-up data: distribution over categories by trialarm (%)


Mobility

Missing 2.9 0.8 1.7

I have no problems in walking about 90.8 93.3 92.2

I have slight problems in walking about 4.0 3.4 3.6

I have moderate problems in walking about 2.3 1.3 1.7

I have severe problems in walking about 0 0.8 0.5

I am unable to walk about 0 0.4 0.2

Self-care

Missing 2.9 0.8 1.7

I have no problems washing or dressing myself 93.7 96.6 95.4

I have slight problems washing ordressing myself

1.7 0.4 1.0

I have moderate problems washing ordressing myself

1.2 1.7 1.5

I have severe problems washing ordressing myself

0.6 0.4 0.5

I am unable to wash or dress myself 0 0 0

Usual activities

Missing 2.9 0.8 1.7

I have no problems doing my usual activities 90.8 90.8 90.8

I have slight problems doing my usual activities 3.5 6.3 5.1

I have moderate problems doing myusual activities

2.9 0.8 1.7

I have severe problems doing my usual activities 0 0.8 0.5

I am unable to do my usual activities 0 0.4 0.2

Pain/discomfort

Missing 3.5 0.8 1.9

I have no pain or discomfort 79.9 80.3 80.1

I have slight pain or discomfort 11.5 13.9 12.9

I have moderate pain or discomfort 4.0 3.8 3.9

I have severe pain or discomfort 1.2 0.8 1.0

I have extreme pain or discomfort 0 0.4 0.2

Anxiety/depression

Missing 2.9 1.3 1.9

I am not anxious or depressed 74.7 64.3 68.7

I am slightly anxious or depressed 8.6 17.2 13.6

I am moderately anxious or depressed 6.3 13.5 10.4

I am severely anxious or depressed 5.8 3.4 4.4

I am extremely anxious or depressed 1.7 0.4 0.9

continued



61

Pilot intermediate (outcome) measures

Participants were asked to complete items addressing attitudinal and knowledge-based precursors tosmoking, including perceived prevalence of smoking (perceived norms), at baseline. The aim was toexplore potential changes in these individual-level characteristics that were targeted via the interventionand explicit in the logic model. NatCen attitudinal scales41 and ESFA items assessing social and situationalself-efficacy and skills42,43 were used in this study. Table 15 shows the distribution of responses by armat baseline.

Participants completed a follow-up survey on attitudinal and knowledge-based precursors to smoking,including perceived prevalence of smoking (perceived norms). The survey tool was adapted from NatCenitems41 social and situational self-efficacy and skills, using the ESFA items.42,43 At follow-up, the surveyincorporated questions on institutional-level influences, including awareness of college practices onsmoking and The Filter project. It also considered community-level data to understand participantexperience of tobacco purchasing and local retailer behaviour. Table 16 shows distribution of responses byarm at follow-up. Table 16 illustrates that a majority of students who attempted to purchase tobacco wereable to do so. Only 5.1% of students were aware of The Filter project at follow-up, although this washigher in the intervention group (7.1%) than in the control group (2.9%).

TABLE 13 Categorical EQ-5D-5L items and health service use at follow-up (continued )

Variable



During the last 30 days, have you visited a doctoror nurse?

Missing 3.5 0.8 1.9

Yes 23.6 23.1 23.3

During the last 30 days, have you been admittedto hospital?

Missing 2.9 0.8 1.7

Yes 6.9 5.0 5.8

How many nights admitted to hospital in last30 days?a

Missing 0 0 0

Day case 83.3 83.3 83.3

1 8.3 8.3 8.3

2 8.3 8.3 8.3

≥ 3 0 0 0

a Information only recorded on those who had been admitted to hospital in last 30 days.

RESULTS


62

TABLE 14 Numerical EQ-5D-5L items and health service use at follow-up


1-year follow-up data: characteristics by trial arm (%)


How good or bad is your health TODAY?[0 (worst you can imagine) to 100 (best you can image)]

Control 168 3.0 10.0 70.1 84.0 90.1 100.0 80.5 16.5

Intervention 236 0.8 8.0 75.0 85.0 93.1 100.0 83.0 14.3

Overall 404 1.9 8.0 75.0 85.0 92.0 100.0 82.0 15.3

How many visits to a doctor or nurse in last 30 days?a (0–6) Control 41 0 1.0 1.0 1.0 2.0 6.0 1.8 1.3

Intervention 55 0 1.0 1.0 1.0 2.0 5.0 1.7 1.1

Overall 96 0 1.0 1.0 1.0 2.0 6.0 1.8 1.2

EQ-5D-5L index score (–1.0 to 1.0) Control 174 0.0 0.3 0.9 1.0 1.0 1.0 0.9 0.1

Intervention 238 0.0 0.1 0.9 1.0 1.0 1.0 0.9 0.1

Overall 412 0.0 0.1 0.9 1.0 1.0 1.0 0.9 0.1

a Information only recorded on those who had been visited a doctor or nurse in last 30 days.

DOI:10.3310/phr05080

PUBLIC

HEA

LTHRESEA

RCH2017

VOL.5

NO.8

©Queen

’sPrinter

andController

ofHMSO

2017.This

work

was

producedby

Fletcheret

al.under

theterm

sof

acom

missioning

contractissued

bythe

Secretaryof

Statefor

Health.

Thisissue

may

befreely

reproducedfor

thepurposes

ofprivate

researchand

studyand

extracts(or

indeed,the

fullreport)may

beincluded

inprofessionaljournals

providedthat

suitableacknow

ledgement

ismade

andthe

reproductionisnot

associatedwith

anyform

ofadvertising.

Applications

forcom

mercialreproduction

shouldbe

addressedto:

NIHRJournals

Library,NationalInstitute

forHealth

Research,Evaluation,

Trialsand

StudiesCoordinating

Centre,

Alpha

House,

University

ofSoutham

ptonScience

Park,Southam

ptonSO

167N

S,UK.

63

TABLE 15 Pilot intermediate outcome measures at baseline

Variable



People of my age smoke because it helps themto relax

Missing 2.3 2.9 2.7

True 60.0 61.8 61.1

False 37.7 35.3 36.3

People of my age smoke because they are addictedto tobacco

Missing 2.6 2.7 2.7

True 77.8 77.3 77.5

False 19.6 20.0 19.8

People of my age smoke because they believe ithelps them to stay slim

Missing 3.2 3.0 3.1

True 27.0 27.3 27.2

False 69.8 69.7 69.7

People of my age smoke because it helps themwith stress in their life

Missing 3.0 3.0 3.0

True 77.7 77.3 77.4

False 19.4 19.7 19.6

People of my age smoke to look cool in front oftheir friends

Missing 2.6 2.7 2.7

True 77.2 77.7 77.5

False 20.2 19.6 19.8

People of my age smoke because they find itexciting to break the rules

Missing 3.0 3.0 3.0

True 52.6 52.6 52.6

False 44.5 44.4 44.4

People of my age smoke because their friendspressure them into it

Missing 2.6 3.0 2.8

True 74.0 72.1 72.9

False 23.4 24.9 24.3

People of my age smoke because it gives them agood feeling

Missing 3.0 3.3 3.2

True 66.2 63.2 64.4

False 30.9 33.5 32.4

RESULTS


64

TABLE 15 Pilot intermediate outcome measures at baseline (continued )

Variable



Have you ever felt pressure to smoke from yourbest friend?

Missing 2.6 3.0 2.8

Often 3.6 1.5 2.4

Sometimes 13.0 10.2 11.3

Never 77.5 81.1 79.6

I do not have one/any 3.4 4.2 3.9

Have you ever felt pressure to smoke fromother friends?

Missing 3.0 3.2 3.1

Often 4.3 2.1 3.0

Sometimes 20.4 18.0 19.0

Never 69.8 73.5 72.0


TABLE 16 Pilot intermediate outcome measures at follow-up

Variable



Individual level

People of my age smoke because it helps themto relax

Missing 2.9 0.8 1.7

True 63.8 73.1 69.2

False 33.3 26.1 29.1

People of my age smoke because they are addictedto tobacco

Missing 3.5 0.8 1.9

True 77.0 83.6 80.8

False 19.5 15.6 17.2

People of my age smoke because they believe ithelps them to stay slim

Missing 2.9 1.3 1.9

True 28.7 28.2 28.4

False 68.4 70.6 70.0

continued



65

TABLE 16 Pilot intermediate outcome measures at follow-up (continued )

Variable



People of my age smoke because it helps themwith stress in their life

Missing 2.9 0.8 1.7

True 80.5 86.6 84.0

False 16.7 12.6 14.3

People of my age smoke to look cool in front oftheir friends

Missing 2.9 0.8 1.7

True 81.0 80.7 80.8

False 16.1 18.5 17.5

People of my age smoke because they find itexciting to break the rules

Missing 2.9 0.8 1.7

True 60.9 60.0 59.2

False 36.2 41.2 39.1

People of my age smoke because their friendspressure them into it

Missing 2.9 0.8 1.7

True 71.8 71.9 71.8

False 25.3 27.3 26.5

People of my age smoke because it gives them agood feeling

Missing 2.9 0.8 1.7

True 60.3 71.0 66.5

False 36.8 28.2 31.8

Have you ever felt pressure to smoke from yourbest friend?

Missing 2.9 1.3 1.9

Often 3.5 1.7 2.4

Sometimes 13.2 10.1 11.4

Never 78.2 84.8 82.0


Have you ever felt pressure to smoke fromother friends?

Missing 2.9 1.3 1.9

Often 3.5 2.5 2.9

Sometimes 16.1 14.3 15.1

Never 76.4 80.7 78.9


RESULTS


66


Variable



Institutional level

Do you ever see students smoking at your college?

Missing 1.7 0.8 1.2

Never 6.3 1.3 3.4

Once or twice a year 2.3 0.8 1.5

Yes, every month 1.7 0.0 0.7

Yes, every week 9.2 6.3 7.5

Yes, daily 78.7 90.8 85.7

Do you think your college could do more toprevent or restrict smoking to certain areas on site?

Missing 2.3 0.8 1.5

Yes 54.6 47.5 50.5

No 27.0 33.6 30.8

Not sure 16.1 18.1 17.2

Does your college have a designated smoking areafor students to smoke?

Missing 2.3 0.8 1.5

Yes 68.9 76.5 73.3

No 23.6 15.6 18.9

Not sure 5.2 7.1 6.3

Do you ever see staff smoking at your college?

Missing 2.9 0.8 1.7

Never 47.7 62.2 56.1

Once or twice a year 8.6 8.4 8.5

Yes, every month 4.0 2.9 3.4

Yes, every week 14.9 9.7 11.9

Yes, daily 21.8 16.0 18.5

Do you think that the staff at your college areconcerned about students smoking?

Missing 2.3 0.8 1.5

Yes 40.2 38.7 39.3

No 33.3 37.4 35.7

Not sure 24.1 23.1 23.5

Do you think that the staff at your college areworking to prevent students from takingup smoking?

Missing 2.3 0.8 1.5

Yes 44.3 31.9 37.1

continued



67


Variable



No 28.2 44.9 37.8

Not sure 25.3 22.3 23.5

Do staff ever warn students about the health risksof smoking?

Missing 2.3 0.8 1.5

Yes 62.1 60.1 60.9

No 24.1 26.5 25.5

Not sure 11.5 12.6 12.1

Do most staff appear confident when discussing asmoking-related issue with students?

Missing 2.3 0.8 1.5

Yes 60.3 60.9 60.7

No 8.1 10.8 9.2

Not sure 29.3 28.2 28.6

Do staff encourage students to use e-cigarettes tohelp them smoke less at college?

Missing 2.3 0.8 1.5

Yes 12.1 5.9 8.5

No 58.1 65.6 62.4

Not sure 27.6 27.7 27.7

Have you heard of The Filter youth project?

Missing 4.0 0.8 2.2

Yes 2.9 7.1 5.3

No 86.8 87.4 87.1

Not sure 6.3 4.6 5.3

Community level

In the past year have you ever gone into a shopnear your college to buy cigarettes or tobacco?(This includes buying for somebody else)

Missing 1.7 0.8

Yes 14.4 12.2

No 83.9 87.0

The last time you went into a shop near yourcollege to buy tobacco or cigarettes, whathappened?a

Missing 0.0 0.0

I bought some cigarettes 96.0 93.1

They refused to sell me cigarettes 4.0 6.9

a Information only recorded on those who had bought cigarettes or tobacco in a shop near their college.

RESULTS


68

Chapter 4 Discussion

This section starts by considering the limitations of this study. First, the limitations of the recruitmentmethods and measures used to survey students about smoking and other health outcomes are discussed,

including sampling and retention problems. Second, other practical and methodological limitations arediscussed, including problems identified with the focus group data collection methods and the deviationsfrom the study protocol. The discussion then focuses on key results from the process evaluation, beforeconsideration of generalisability and the implications of findings for intervention development in FE settingsand future research are discussed. The final section reports conclusions and recommendations.

Limitations

Student survey limitationsAlthough data indicate acceptability of survey methods to participants, with very low levels of missing dataat both baseline and follow-up, the utility of this survey approach was compromised by issues of accessand recruitment highlighted in Chapter 3, Research methods: feasibility and acceptability. A key limitationwith the survey methods was that establishing accurate data on total eligible sample size at each college atbaseline was problematic because of a lack of accurate, up-to-date enrolment data. These issues wereespecially prevalent in large FE settings as a result of students enrolling, in principle, prior to September,but not registering at the start of term; students deferring or dropping out in early September; inclusion ofstudents who study across multiple campuses but whose primary campus is not the study site; and theinclusion of some students aged > 18 years as a result of incomplete information at enrolment. This meantthat the number of potentially eligible students provided by colleges included some students who wereeither aged < 16 years or > 18 years and are, therefore, ineligible.

As accurate information on the number of students aged between 16 and 18 years who were attendingcolleges was not available, the denominator included ineligible participants and response rates would havebeen higher if they were not included. Ineligible participants were older, worked part time and were morelikely to be classified as a weekly smoker than those who were eligible. Initial data provided by collegesdid not – and could not – account for changes attributable to early exit by students or course changes.Although the intervention was specifically aimed at new students, commencement of the study at the startof the academic year emerged as problematic for student engagement, as staff did not yet know studentsor tutor group composition. Beginning the study further into the academic year would not likely haveaided recruitment, with many students out on work placements and attending sporadically once coursesare fully under way.

Although response rates reported at baseline are underestimates, overall the response and retention ratesin this study would be too low to power a cluster RCT design aiming to assess effectiveness, and it isuncertain whether or not those students undertaking the survey are representative of the student body asa whole. Attendance patterns further impacted data collection as students are not in set places and atset times, meaning no fixed data collection points, with researchers instead relying on passing traffic.Although access to tutor groups would appear to present a partial solution to these issues, lack of accessto these groups was common.

Respondents in the intervention and control groups were not equivalent at baseline in terms of age orgender, which may have impacted outcomes if baseline smoking attitudes and behaviour differed. The trialarms were not well balanced for the indicative primary and secondary outcome measures at baseline orfollow-up. This imbalance is not unusual given that there were few clusters and heterogeneity betweenclusters, and would not be considered a risk in a larger cluster RCT assessing effectiveness.



69

Although baseline survey participation rates were reasonable, the research experienced high levels ofattrition between baseline and follow-up. Thirty-six per cent of baseline respondents completed the 1-yearfollow-up survey, constituting 17% of potentially eligible participants. This low response at follow-up shouldbe considered in interpreting results and, for future research, analysis of characteristics of non-responders atfollow-up is recommended to understand how this may be better addressed.

Response rates at follow-up were impacted by a lack of effective channels to contact students, includinglimited access to personal e-mail, and student reluctance to respond to unsolicited telephone calls.Although it may have seemed reasonable to surmise that FE students would constitute a seemingly captiveaudience similar to that in schools, in practice this was not the case. The identification of multipleproblems in recruitment and retention was evident in follow-up response rates, which indicate a lack offeasibility for retention in this setting. Although it was assumed that existing social media channels wouldbe a useful tool for promotion of – and retention in – the trial, in practice, levels of engagement withthese channels were low. It cannot be established whether this is typical of FE settings or specific to thesesample sites, suggesting that this should be considered for future research through pre-interventiondiscussion with staff and students.

In relation to selection of survey tools, both the EQ-5D-5L scale and the health service use measures hadlimitations that should be considered for future research. As the age group under investigation generallyhas low engagement with health services and, because of the limited timescale of the study, it may havebeen unrealistic to expect change when rates of smoking-related health problems are likely to be lowcompared with older smokers. This further suggests that the EQ-5D-5L would not be sensitive enough topick up changes to health or quality of life over a time scale this short, and it is recommended for futurework in this area that other, purpose-built scales are developed that are more sensitive to likely healthimpacts among this age group.

Limitations with other data collection methodsThere are some limitations in our assessment of student and staff views on acceptability associated withsampling and recruitment to focus groups. The process of gathering students and staff members togetherfor focus groups proved time-consuming and challenging. It is uncertain how representative focus groupparticipants were of the wider staff and student populations, and there was limited time to explore thisin detail. No socioeconomic information about the student participants in focus groups was obtained.Difficulties in securing staff time away from routine practice impacted both focus group participation andattendance at staff training, which was delivered under target levels as a result of limited demand by sites.This was attributed by staff to pressures on time and may have been further impacted by the lack of clarityon intervention content indicated in qualitative data obtained from the process evaluation.

Stratification to student focus groups by smoking characteristics was particularly challenging, not leastbecause a key finding of this study (see Chapter 3, Attitudes towards smoking in the further educationcontext) indicates that many FE students who smoke were reluctant to define themselves as ‘smokers’.Although some people who were regular smokers clearly defined themselves as such, the grouping of‘non-smokers’ was, in practice, divided into those who did not smoke at all and those who wereoccasional, or ‘social’ smokers.

Mystery shopper activities were inconsistently delivered and recorded, with variation in how the sameshops were classified at baseline and follow-up; for example, some shops that had been classified asconvenience stores at baseline were classified as confectionery, tobacconist, newsagents at follow-up, andvice versa. This suggests that the definition of some shop types was not clear and further guidance wasrequired. The data for shops classified as convenience stores or confectionery, tobacconist, newsagents aretherefore presented together in Chapter 3. Variance also occurred in how the mystery shopper survey wascompleted between baseline and follow-up, even when the accompanying fieldworker had remained thesame. In particular, the dates and times of visits were not precisely recorded at follow-up (only afternoon/evening stated, rather than the exact time).

DISCUSSION


70

The retailer elements of the study were also impacted by the locations of participating sites and, particularly,the lack of retailers in close proximity to colleges, which may have affected consistency of staff interactionwith shops. The combined limitations of implementation and access mean that the impact of the interventionon local sales practices, particularly underage sales to students aged 16–17 years, cannot be fully assessed atthis time. Future interventions that aim to incorporate a retailer behaviour element should consider this priorto implementation.

Intervention staff checklists were poorly completed throughout the project, which partly reflects high staffturnover within some small, voluntary sector teams such as this. This turnover further impacted FE staffengagement, with frustrations expressed at the number of intervention staff involved and the lack of aconsistent point of contact. For future intervention delivery in this setting, the utilisation of permanentlycontracted staff, if available, should be considered. Implementation may be supported by earlierengagement with FE settings to tailor and deliver the intervention with college managers and other staff.

No major problems were encountered in securing access to conduct observations in the intervention sites;however, fieldworkers at the larger sites reported difficulties in ensuring that all locations had beenchecked thoroughly to identify if messages were consistent and all possible data were collected. The sameresearcher did not always complete the baseline and follow-up observations at each site, so differencesbetween baseline and follow-up data may be because of observer bias. More detailed observations couldbe obtained if fieldworkers were familiar with the site and can visit more than once and at different timesof the day to observe smoking behaviours. A more structured survey to monitor sites at baseline andfollow-up might provide more accurate detail about changes over time. The use of a tablet computer toinput the online survey on site worked well in most cases, although data from one site did not uploadproperly, so the fieldworker completed the survey from memory.

Deviations from the protocolThe only major deviation from the protocol was that post-follow-up saliva testing (cotinine and anabasinetests) did not go ahead as planned because the research team was unable to recruit students to providethese samples. This means that the validity of self-reported smoker status could not be verified in thisstudy. For future research in this setting, prior exploration of the acceptability of saliva sample collectionand testing should be undertaken with students to identify new methods of recruitment and betterincentives that may improve participation.

The study protocol also included plans for economic analyses to include an assessment of the costs of theintervention based on reports of the resources used by ASH Wales’ staff and FE colleges. Within this, keyinterventional resources were considered to include intervention staff time (intervention manager, trainingand education officers, web and social media officers, youth workers), as well as training event and youthworkshop travel and resources, and other consumables relating to the delivery of the project. However,it was not possible to do these planned economic analyses based on the costs of staff time and otherintervention expenses because we did not receive these data from the intervention team.

Key results and generalisability

AcceptabilityThis section considers key findings from the process evaluation, including the influence of non-interventionistinstitutional cultures identified at participating sites and the perceptions of acceptability of interventioncomponents.

Although the study was not powered to detect adverse intervention effects, these were considered inanalysis of qualitative data, with no unintended harms to student participants or staff identified. Theacceptability of the intervention to FE colleges suggests little risk of disruption to practice. However, lack of



71

delivery of the policy review component, and staff disappointment with this, may have affected therelationships between colleges and research teams, with implications for future participation in similar trials.

A significant barrier to acceptability of intervention content, and the willingness of staff to fully engage,relates to culture within the FE sector, particularly large colleges, in which personal choice and freedom arehighly valued and the voluntary nature of attendance is recognised. Both staff and students strongly valuedthe transitional nature of FE as an environment in which development from youth to adulthood occurs,with corresponding emphasis on personal responsibility, including for one’s own smoking behaviour. Thiswas evident in the contrast between support for formal smoke-free norms as embodied in policy in FEsettings, including approval for smokers to be regulated and restricted to approved territories, and theinformal tolerance exhibited in staff attitudes to challenging smoking behaviour and intervening withyoung adult choices.

Although this finding reflects perceptions commonly found in higher education settings of students asadults free to pursue their own choices, in contrast with higher education, students in FE settings may nothave the same legal status. As it is illegal to sell tobacco to those aged < 18 years, this suggests scope toquery the non-interventionist stance commonly expressed. Further research would aid understanding ofhow to balance the necessary emphasis on students’ transitional and developmental needs and the legalframework FE settings operate in. Cross-cultural comparisons with educational systems in whichattendance to the age of 18 years is compulsory may also aid understanding of the role of voluntaryparticipation in FE as a contributor to the development of more permissive environments.

This cultural difference between school and FE was also evident in the view that FE may be ‘too late’ forprevention activities, suggesting a need to further explore smoking cessation in this context. Students andstaff suggested that they had ‘heard it all before’ and were well enough informed on smoking harms,especially non-smokers who did not want to engage with seemingly irrelevant information. The lack ofacceptability of content was evident in low engagement with youth work sessions focused on developmentof smoke-free message campaigns, which were felt to be more relevant to younger age groups, suggestinga more nuanced approach could be needed in this setting. This was recognised by the intervention teamand contrasted with previous experience of delivery in school settings. However, challenging the limitedacceptability for prevention in the setting may be important in light of evidence that 60% of adult smokerscommence regular smoking after the age of 16 years, suggesting an unmet need for prevention aftercompulsory education. Further exploration of prevention delivery in post-16 settings should consider whatmight constitute acceptable intervention content to target audiences.

Data suggest high acceptability among FE staff for intervention activities focused on smoking policy reviewand development but, as a result of intervention management issues, policies were not reviewed asintended. Randomisation was acceptable to FE managers, who also perceived added value in participationthrough engagement with external organisations and student exposure to higher education research. Staffalso indicated acceptability of the training component of the intervention, particularly when they perceivedexisting knowledge gaps, such as with e-cigarettes. However, this was not matched by perceived likelihoodof utilising content in discussing smoking with students, reflecting the lack of support for interventionistapproaches that contrast with the emphasis on personal responsibility.

The intervention team found FE settings challenging and not conducive to smoking prevention activitiescompared with other educational settings (e.g. they reported significant differences between the prevailingculture of FE colleges in this study and schools where they had worked), which was observed as impactingstudent engagement. They also noted limitations in terms of access to target populations in FE settingsattributable to existing organisational behaviours and activities.

ReachData indicate low levels of engagement with social media channels used for message dissemination,reflecting a possible misperception in intervention development, as well as institutional barriers, including

DISCUSSION


72

college variations in the level of staff engagement in social media channels. Although students reportedroutinely accessing social media, this did not seem to include use of formal college websites and socialmedia channels, suggesting overestimation of the likely impact of this approach. This finding is significantfor future intervention delivery in this setting, including informing recruitment and retention strategies thatmay aim to utilise social media.

A further limitation of the intervention – and the logic model – emerged from process data, in which studentsfrequently reported obtaining tobacco from peers and family rather than local retailers, suggesting targetinglocal retailer practice was likely to have limited effect. The majority of students who tried to purchasecigarettes were able to do so successfully at follow-up, suggesting limited penetration of community-levelactivities. The majority of students were unaware of The Filter project at follow-up, suggesting limited impactof awareness-raising activities supporting roll-out, such as social media promotion, as discussed above.

The identification of ‘informal’ supply chains in this study has implications for targeting of any futureinterventions, and further research would help to identify whether this supply was primarily taking tobaccoproducts from inside the home or proxy purchasing (someone aged > 18 years purchasing tobaccoproducts for someone aged < 18 years). Despite the legal age for tobacco purchasing increasing from 16to 18 years in the UK in 2007, the law on illegality of proxy purchasing was only implemented in October2015 after follow-up analysis. It is important to note that retailer behaviour in any future interventions maybe differentially impacted by any enforcement activity associated with this legal change.

Implications

Overall, the lack of support for the intervention, combined with institutional factors that impacted delivery,such as short run-in, FE timetabling and systems, and poor intervention management, meant that The FilterFE was not acceptable or feasible at any of the pilot sites at this time. However, it cannot be concluded fromthis research that individual components of the intervention could not transfer to FE settings effectively ifsome of the key implementation issues identified were addressed in future studies.

This research has identified multiple factors that suggest that these FE settings were not conducivecontexts for the delivery of prevention interventions. The prevailing non-interventionist ethos evident in FEcolleges – more so than school-based sixth forms – suggests a different approach is required in such large,transitional environments to increase acceptability. However, although findings from the process evaluationappear to strongly support resistance to prevention messages, interpretation of the validity of otherintervention components should be assessed in light of implementation issues, such as those impacting theassessment of retailer practices. Although not successful in this study, staff willingness to engage onpolicy-related matters may also suggest that this component would benefit from further investigation andreassessment of impact. It is also noted that FE managers were keen to work with universities and outsideagencies, particularly when knowledge gaps are perceived, as with e-cigarettes, suggesting coproductionin future intervention development to identify issues and approaches deemed pertinent. Although smokingprevention was not a receptive message for this population, interest was further expressed in access tosmoking cessation (i.e. not accessed via general practitioner or health services, services that do not needtransport to get to), as well as policy support for colleges, suggesting scope for future work.

This research illustrates the variations within this sector that make generalising from results problematic. Thediversity of the sector, in terms of mix of socioeconomic groups, vocational and academic courses, and thesettings students are exposed to, including workplaces, requires further enquiry to inform current knowledgeon the impact of FE on health behaviours in young people. Exploratory analysis of organisational features,drawing on research into university culture and processes, could inform future intervention development.The lack of homogeneity across FE illustrates the necessity of flexibility in intervention developmentand delivery.



73

Overall, the limitations to data collection methods, combined with the limited implementation of somecomponents, means that the potential impacts of smoking prevention activities across multiple levels(student, college, community) remain hard to interpret. Findings are most robust at the level of individualinfluence, with this research increasing understanding of issues of reach and acceptability for individuals inthese settings. Organisationally, the process evaluation has highlighted important procedural andattitudinal influences on on-site intervention components. Questions have been raised regarding the roleof organisational processes and norms on implementation of prevention interventions, which can beconsidered for future development work and delivery. At the level of community effects, implementationissues relating to retailer engagement and monitoring mean that little can be asserted about the retailercomponent of the intervention. Although a socioecological approach presented challenges in terms of therange and complexity of intervention components involved, such a broad exploration was beneficial inhighlighting the range of issues impacting smoking behaviour and intervention delivery in this setting.

Conclusions and recommendations for further research

The principal conclusion and recommendation of this report is for no further evaluation of this smokingprevention intervention in FE settings without consideration of how to address the issues identified inimplementation of intervention components and acceptability of this type of intervention. However, even ifsome practical delivery constraints could be overcome in future interventions, issues of acceptability andreach may still occur in this setting due to relatively deep-rooted cultural and procedural constraintsreported. The process evaluation data revealed how institutional cultures within FE settings limit theacceptability of prevention activities for smoking, although it is recommended that this be investigatedacross a broader range of FE schools and colleges, including those in denser urban areas and with othersociodemographics. It is also recommended that more information on socioeconomic status be capturedfor future studies in FE settings to better understand potential contextual influences and barriers.

We would recommend that other pilot trials and process evaluations of pilot interventions adopt amixed-methods, socioecological approach. Although there were various limitations with our data collectionmethods, these were partially counteracted by the use of multiple data sources to aid triangulation andfieldworker immersion to enhance depth of understanding, as well as robust review methods among thestaff team. This included development of a framework for assessing individual results against RQs andgroup discussion of identified contradiction in the data. The strength of multiple data collection wasfurther enhanced by the adoption of a socioecological lens, which ensured consideration of influencesacross individual, institutional and community levels. This ensured that data saturation was probablyreached, with some inductive themes emerging via the data that could be researched further (e.g. peerinfluence and the role of institutional culture).

Staff and student views suggest that smoking cessation activities may be appropriate and acceptable inthis setting, and it is suggested that these should be a focus for further intervention development research.Should further research and development take place in FE colleges, key considerations would includetiming of engagement and intervention delivery of smoking cessation activities. Overall, this study suggeststhat there will be greater value in smoking cessation intervention activities than smoking preventionactivities. This is because of the relatively low numbers of students who reported taking up smoking in thefirst year of their FE courses combined with the low levels of acceptability regarding prevention messagesin this context.

It was feasible to recruit, randomise and retain FE settings within a cluster RCT design. FE managers valuedthe opportunity to be involved in health research, particularly the input of external agencies and theadditional resources provided, and accepted randomisation. However, further methodological work isrecommended to improve student recruitment and retention rates prior to any larger RCTs beingconducted in this setting. Problems with identification of the baseline population would suggest potentialdelivery later in the academic year when the student population has stabilised. However, differences

DISCUSSION


74

between school- and college-based FE settings are important, with more vocational colleges likely to havemore sporadic campus attendance than more academic settings, suggesting a need for novel approachesto student engagement and recruitment.

Finally, this study benefited from extensive patient and public involvement with young people through theAdvice Leading to Public Health Advancement (ALPHA) youth group. We would recommend furtherpatient and public involvement with young people to help steer more successful evaluation design anddevelop trial methods in future research. For example, young people should be consulted prior to newresearch to develop new interventions targeting cessation as well as prevention, or new research todevelop methods for recruiting and retaining more young people in FE settings.



75

Acknowledgements

The work was undertaken with the support of The Centre for the Development and Evaluation ofComplex Interventions for Public Health Improvement, a UK Clinical Research Collaboration PHR Centre

of Excellence. Joint funding (MR/KO232331/1) from the British Heart Foundation, Cancer Research UK,Economic and Social Research Council, Medical Research Council, the Welsh Government and theWellcome Trust, under the auspices of the UK Clinical Research Collaboration, is gratefully acknowledged.

Contributions of authors

Adam Fletcher planned and managed the project, including management of baseline survey collection,and led on development of data collection tools, the synthesis of quantitative and qualitative data sourcesto assess progression criteria, and reporting.

Micky Willmott led on the analysis and write-up of qualitative data, and contributed towards to thesynthesis of quantitative and qualitative data sources to assess progression criteria, and report drafting.

Rebecca Langford managed the process evaluation, led on the collection and analysis of the qualitativedata, consultation with young people and write-up of process evaluation findings.

James White conducted the statistical analysis, led on the drafting of the statistical findings sections andcontributed to drafting other sections of the report.

Ria Poole managed the collection and analysis of follow-up survey data and contributed towardsreport drafting.

Rachel Brown provided support to the team on drafting and editing the report.

Honor Young contributed to the collection of baseline and follow-up survey data, and commented onreport drafts.

Graham Moore contributed to project planning, advised on pilot outcomes and commented onreport drafts.

Simon Murphy contributed to project planning and commented on report drafts.

Julia Townson contributed to project planning and commented on report drafts.

William Hollingworth advised on the economic evaluation and commented on report drafts.

Rona Campbell contributed to project planning and commented on report drafts.

Chris Bonell supported the design of the process evaluation and contributed to project planning andreport drafting.

Data sharing statement

We shall make data available to the scientific community with as few restrictions as feasible. Please contactthe corresponding author if you would like to access the data.



77

References

1. Backinger CL, Fagan P, Matthews E, Grana R. Adolescent and young adult tobacco prevention andcessation: current status and future directions. Tob Control 2003;12(Suppl. 4):46–53. https://doi.org/10.1136/tc.12.suppl_4.iv46

2. Office for National Statistics (ONS). General Lifestyle Survey Overview: A Report on the 2010General Lifestyle Survey. Newport: ONS; 2010.

3. National Institute for Health and Care Excellence (NICE). School-based Interventions to PreventSmoking. NICE Public Health Guidance 23. London: NICE; 2010.

4. Bonell C, Jamal F, Harden A, Wells H, Parry W, Fletcher A, et al. Systematic review of the effectsof schools and school environment interventions on health: evidence mapping and synthesis.Public Health Res 2013;1(1).

5. Langford R, Bonell CP, Jones HE, Pouliou T, Murphy SM, Waters E, et al. The WHO HealthPromoting School framework for improving the health and well-being of students and theiracademic achievement. Cochrane Database Syst Rev 2014;4:CD008958. https://doi.org/10.1002/14651858.CD008958.pub2

6. O’Donnell T, Gray G. The Health Promoting College. London: Health Education Authority; 1993.

7. James K. A Health Promoting College for 16–19 Year Old Learners. London: Department ofHealth; 2003.

8. McCambridge J, Strang J. The efficacy of single-session motivational interviewing in reducing drugconsumption and perceptions of drug-related risk and harm among young people: results from amulti-site cluster randomized trial. Addiction 2004;99:39–52. https://doi.org/10.1111/j.1360-0443.2004.00564.x

9. Drury C, Doherty S. An Audit of Healthy School Programmes Working With Colleges in England.Preston: Healthy Settings Development Unit, University of Central Lancashire; 2005.

10. Doherty S, Dooris M. The healthy settings approach: the growing interest within colleges anduniversities. Educ Health 2006;24:42–3.

11. Newbery N, McCambridge J, Strang J. Let’s talk about drugs: pilot study of a community-level drugprevention intervention based on motivational interviewing principles. Health Educ J 2007;107:276–89.https://doi.org/10.1108/09654280710742573

12. Slym RL, Day M, McCambridge J. National survey of drugs and alcohol provisions within furthereducation colleges in England in 2006. Drugs Educ Prev Polic 2007;14:193–204. https://doi.org/10.1080/09687630601132702

13. Twardzicki M. Challenging stigma around mental illness and promoting social inclusion usingthe performing arts. J R Soc Promot Health 2008;128:68–72. https://doi.org/10.1177/1466424007087804

14. Warwick I, Statham J, Aggleton P. Healthy and Health Promoting Colleges: Identifying an EvidenceBase. London: Thomas Coram Research Unit, Institute of Education, University of London; 2008.

15. Warwick I, Maxwell C, Statham J, Aggleton P, Simon A. Supporting mental health and emotionalwell-being among younger students in further education. JFHE 2008;32:1–13. https://doi.org/10.1080/03098770701560331

16. Richards C. How Useful are Bounded Online Chat Rooms as a Source of Pastoral Support in aSixth-Form College? PhD thesis. Southampton: University of Southampton, School of Education; 2009.



79

https://doi.org/10.1136/tc.12.suppl_4.iv46

https://doi.org/10.1136/tc.12.suppl_4.iv46

https://doi.org/10.1002/14651858.CD008958.pub2


https://doi.org/10.1111/j.1360-0443.2004.00564.x

https://doi.org/10.1111/j.1360-0443.2004.00564.x

https://doi.org/10.1108/09654280710742573

https://doi.org/10.1080/09687630601132702

https://doi.org/10.1080/09687630601132702

https://doi.org/10.1177/1466424007087804

https://doi.org/10.1177/1466424007087804

https://doi.org/10.1080/03098770701560331

https://doi.org/10.1080/03098770701560331

17. McCambridge J, Hunt C, Jenkins RJ, Strang J. Cluster randomised trial of the effectiveness ofmotivational interviewing for universal prevention. Drug Alcohol Depend 2011;114:177–84.https://doi.org/10.1016/j.drugalcdep.2010.07.028

18. Audrey S, Langford R. Dying to get out: young drivers, safety and social inequity. Inj Prev2014;20:1–6. https://doi.org/10.1136/injuryprev-2013-040756

19. Minary L, Cambon L, Martini H, Wirth N, Acouetey DS, Thouvenot F, et al. Efficacy of a smokingcessation program in a population of adolescent smokers in vocational schools: a public healthevaluative controlled study. BMC Public Health 2013;13:149. https://doi.org/10.1186/1471-2458-13-149

20. Stead LF, Lancaster T. Interventions for preventing tobacco sales to minors. Cochrane Database SystRev 2008;1:CD001497.

21. Brinn MP, Carson KV, Estermann AJ, Chang AB, Smith BJ. Mass media interventions for preventingsmoking in young people. Cochrane Database Syst Rev 2010;11:CD001006. https://doi.org/10.1002/14651858.CD001006.pub2

22. Hopkins DP, Razi S, Leeks KD, Priya Kalra G, Chattopadhyay SK, Soler RE, Task Force onCommunity Preventive Services. Smokefree policies to reduce tobacco use. A systematic review.Am J Prev Med 2010;38(Suppl. 2):275–89. https://doi.org/10.1016/j.amepre.2009.10.029

23. Carson KV, Brinn MP, Labiszewski NA, Esterman AJ, Chang AB, Smith BJ. Community interventionsfor preventing smoking in young people. Cochrane Database Sys Rev 2011;7:CD001291.https://doi.org/10.1002/14651858.CD001291.pub2

24. Thomas RE, McLellan J, Perera R. School-based programmes for preventing smoking. CochraneDatabase Syst Rev 2013;4:CD001293. https://doi.org/10.1002/14651858.CD001293.pub3

25. Oude Luttikhuis H, Baur L, Jansen H, Shrewsbury VA, O’Malley C, Stolk RP, et al. Interventions fortreating obesity in children. Cochrane Database Syst Rev 2009;1:CD001872. https://doi.org/10.1002/14651858.CD001872.pub2

26. Greaves CJ, Sheppard KE, Abraham C, Hardeman W, Roden M, Evans PH, et al. Systematic reviewof reviews of intervention components associated with increased effectiveness in dietary andphysical activity interventions. BMC Public Health 2011;11:119. https://doi.org/10.1186/1471-2458-11-119

27. Chokshi DA, Farley TA. The cost-effectiveness of environmental approaches to disease prevention.N Engl J Med 2012;367:295–7. https://doi.org/10.1056/NEJMp1206268

28. White M, Adams J, Heywood P. How and Why do Interventions that Increase Health Overall WidenInequalities within Populations? In Babones SJ, editor. Social Inequality and Public Health. Bristol:Policy Press; 2012. pp. 65–82.

29. Lorenc T, Petticrew M, Welch V, Tugwell P. What types of interventions generate inequalities?Evidence from systematic reviews. J Epidemiol Community Health 2013;67:190–3. https://doi.org/10.1136/jech-2012-201257

30. Glasgow RE, Lichtenstein E, Marcus AC. Why don’t we see more translation of health promotionresearch to practice? Rethinking the efficacy-to-effectiveness transition. Am J Public Health2003;93:1261–7. https://doi.org/10.2105/AJPH.93.8.1261

31. McLeroy KR, Bibeau D, Steckler A, Glanz K. An ecological perspective on health promotionprograms. Health Educ Q 1988;15:351–77. https://doi.org/10.1177/109019818801500401

32. Durlak JA, DuPre EP. Implementation matters: a review of research on the influence ofimplementation on program outcomes and the factors affecting implementation. Am J CommunityPsychol 2008;41:327–50. https://doi.org/10.1007/s10464-008-9165-0

REFERENCES


80

https://doi.org/10.1016/j.drugalcdep.2010.07.028

https://doi.org/10.1136/injuryprev-2013-040756

https://doi.org/10.1186/1471-2458-13-149

https://doi.org/10.1186/1471-2458-13-149



https://doi.org/10.1016/j.amepre.2009.10.029





https://doi.org/10.1186/1471-2458-11-119

https://doi.org/10.1186/1471-2458-11-119

https://doi.org/10.1056/NEJMp1206268

https://doi.org/10.1136/jech-2012-201257

https://doi.org/10.1136/jech-2012-201257

https://doi.org/10.2105/AJPH.93.8.1261

https://doi.org/10.1177/109019818801500401

https://doi.org/10.1007/s10464-008-9165-0

33. Hawe P, Shiell A, Riley T. Theorising interventions as events in systems. Am J Community Psychol2009;43:267–76. https://doi.org/10.1007/s10464-009-9229-9

34. Action on Smoking and Health (ASH). Australia Guidance for a Tobacco Free Campus: Why WeShould Make Campuses Tobacco Free and How to Do It. Sydney, NSW: ASH Australia; 2009.

35. Fletcher A, Jamal F, Moore G, Evans R, Murphy S, Bonell C. Realist complex intervention science:applying realist principles across all phases of the MRC framework for developing and evaluatingcomplex interventions. Evaluation 2016;22:286–303. https://doi.org/10.1177/1356389016652743

36. Craig P, Dieppe P, Macintyre S, Michie S, Nazareth I, Petticrew M. Developing and evaluatingcomplex interventions: the new Medical Research Council guidance. BMJ 2008;337:a1655.https://doi.org/10.1136/bmj.a1655

37. European Monitoring Centre for Drugs and Drug Addiction (EMCDDA). Handbook for Surveys onDrug Use among the General Population. Lisbon: EMCDDA; 2002.

38. Bush K, Kivlahan DR, McDonell MB, Fihn SD, Bradley KA. The AUDIT alcohol consumptionquestions (AUDIT-C): an effective brief screening test for problem drinking. Ambulatory CareQuality Improvement Project (ACQUIP). Alcohol Use Disorders Identification Test. Arch Intern Med1998;158:1789–95. https://doi.org/10.1001/archinte.158.16.1789

39. Herdman M, Gudex C, Lloyd A, Janssen M, Kind P, Parkin D, et al. Development and preliminarytesting of the new five-level version of EQ-5D (EQ-5D-5L). Qual Life Res 2011;20:1727–36.https://doi.org/10.1007/s11136-011-9903-x

40. Borland R, Yong HH, O’Connor RJ, Hyland A, Thompson ME. The reliability and predictive validityof the Heaviness of Smoking Index and its two components: findings from the InternationalTobacco Control Four Country study. Nicotine Tob Res 2010;12:S45–50. https://doi.org/10.1093/ntr/ntq038

41. Fuller E. Survey of Smoking, Drinking and Drug Use among Young People in England in 2013.London: NatCen; 2013.

42. de Vries H, Mudde A, Kremers S, Wetzels J, Uiters E, Ariza C, et al. The European SmokingPrevention Framework Approach (ESFA): short-term effects. Health Educ Res 2003;18:649–63.https://doi.org/10.1093/her/cyg033

43. de Vries H, Dijk F, Wetzels J, Mudde A, Kremers S, Ariza C, et al. The European Smokingprevention Framework Approach (ESFA): effects after 24 and 30 months. Health Educ Res2006;21:116–32. https://doi.org/10.1093/her/cyh048

44. Hill M. Ethical Considerations in Researching Children’s Experiences. In Greene S Hogan D, editors.Researching Children’s Experience. London: Sage Publications; 2005. pp. 61–86.

45. Knight JR, Sherritt L, Harris SK, Gates EC, Chang G. Validity of brief alcohol screening tests amongadolescents: a comparison of the AUDIT, POSIT, CAGE, and CRAFFT. Alcohol Clin Exp Res2003;27:67–73. https://doi.org/10.1111/j.1530-0277.2003.tb02723.x



81

https://doi.org/10.1007/s10464-009-9229-9

https://doi.org/10.1177/1356389016652743

https://doi.org/10.1136/bmj.a1655

https://doi.org/10.1001/archinte.158.16.1789

https://doi.org/10.1007/s11136-011-9903-x

https://doi.org/10.1093/ntr/ntq038

https://doi.org/10.1093/ntr/ntq038

https://doi.org/10.1093/her/cyg033

https://doi.org/10.1093/her/cyh048

https://doi.org/10.1111/j.1530-0277.2003.tb02723.x

Part of the NIHR Journals Library www.journalslibrary.nihr.ac.uk

Published by the NIHR Journals Library

This report presents independent research funded by the National Institute for Health Research (NIHR). The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health

EMEHS&DRHTAPGfARPHR

further education settings. Public Health Research 5(8 ... · Pilot trial and process evaluation of...

Documents

Transcript of further education settings. Public Health Research 5(8 ... · Pilot trial and process evaluation of...